Intelligent monitoring method and system for energy consumption and carbon emission of data center

Abstract

The invention discloses an intelligent monitoring method and system for energy consumption and carbon emission of a data center, and the method comprises the steps: monitoring energy consumption monitoring data and environment data in an operation environment of the data center, recognizing an energy consumption mode through time sequence analysis, and generating grading quality configuration; executing marginal utility evaluation based on an energy consumption mode to determine a precision configuration boundary, generating a carbon source intensity index through influence intensity evaluation, and converting the carbon source intensity index into an influence coverage area; analyzing the change track of the environmental data to construct a carbon propagation network, acquiring a key propagation path and reconstructing a monitoring network; evaluating an abnormal propagation range influencing the coverage area, determining a monitoring extension area, constructing adaptive monitoring configuration, and identifying an emission reduction potential area; an emission reduction potential area is decomposed into space monitoring nodes, utility weight distribution is carried out, matrix operation is executed through differential monitoring sequences to generate an energy consumption and carbon emission analysis matrix, finally monitoring configuration parameters are generated through statistical evaluation, and intelligent monitoring management of energy consumption and carbon emission of the data center is achieved.

Description

Technical Field

[0001] This invention relates to the field of energy management technology, and in particular to a method and system for intelligent monitoring of energy consumption and carbon emissions in data centers. Background Technology

[0002] Against the backdrop of accelerated digital transformation, data centers, as key infrastructure supporting cloud computing, artificial intelligence, and big data applications, are exhibiting a trend towards large-scale and intensive development. With the booming global digital economy, the energy consumption and carbon emission intensity of data centers are constantly rising, and their environmental impact has attracted widespread international attention.

[0003] Traditional data center energy consumption and carbon emission monitoring technologies primarily rely on fixed monitoring point layouts and static threshold control strategies, which have significant limitations in practical applications. Existing monitoring systems typically employ uniform monitoring accuracy standards, failing to differentiate configurations based on different equipment types and regional importance, resulting in unreasonable allocation of monitoring resources. Traditional analysis methods are mostly based on simple statistical processing of historical data, lacking in-depth analysis of energy consumption change patterns and spatial distribution patterns, making it difficult to accurately identify energy consumption anomalies and carbon emission hotspots. Current technologies lag behind in monitoring network layout optimization; monitoring point configurations are mainly based on experience-based judgments, lacking scientific layout optimization methods and dynamic adjustment mechanisms, and thus unable to effectively adapt to changes in data center equipment configurations and fluctuations in business load. Summary of the Invention

[0004] This invention provides a method and system for intelligent monitoring of energy consumption and carbon emissions in data centers. By deeply analyzing the correlation characteristics between environmental parameters and energy consumption monitoring data, it constructs a precision configuration boundary and a carbon source intensity analysis system, generates a carbon propagation network and identifies key propagation paths, realizes dynamic range adjustment and adaptive monitoring configuration of the monitoring extension area, and finally generates differentiated monitoring sequences and energy consumption and carbon emissions analysis results through the optimization of the utility weights of spatial monitoring nodes and matrix operation analysis, providing intelligent and refined energy consumption and carbon emissions monitoring and management solutions for data centers.

[0005] The first aspect of this invention proposes a smart monitoring method for energy consumption and carbon emissions in data centers, comprising the following steps: Monitor energy consumption data and environmental parameter data within the data center operating environment, perform time-series analysis on the energy consumption data to identify energy consumption patterns, and generate graded quality parameters based on the quality requirements of the energy consumption data; Based on the energy consumption pattern and the graded quality parameters, a marginal utility assessment is performed to determine the precision configuration boundary. An impact intensity assessment is then performed on the precision configuration boundary to generate a carbon source intensity index. Spatial mapping processing is then used to convert the carbon source intensity index into an impact coverage area. The change trajectory of the environmental parameter data is analyzed to generate a carbon propagation network. Frequency analysis is performed on the carbon propagation network to obtain key propagation paths. The key propagation paths are mapped to carbon mobility indicators. Based on the carbon mobility indicators, the monitoring network is reconstructed to form a path optimization table. Assess the abnormal propagation range of the affected coverage area to determine the monitoring extension area, perform dynamic range adjustment on the monitoring extension area to generate range adjustment parameters, fuse the range adjustment parameters with the path optimization table to construct an adaptive monitoring configuration, and identify emission reduction potential areas based on the adaptive monitoring configuration; The emission reduction potential area is decomposed into spatial monitoring nodes. The spatial monitoring nodes are assigned utility weights to generate differentiated monitoring sequences. Matrix operations are performed on the differentiated monitoring sequences to generate an energy consumption and carbon emission analysis matrix. The energy consumption and carbon emission analysis matrix is ​​statistically evaluated to generate monitoring configuration parameters, thus completing the intelligent monitoring of energy consumption and carbon emission in the data center.

[0006] A second aspect of this invention provides a smart monitoring system for energy consumption and carbon emissions in data centers, comprising: The data acquisition module is used to monitor energy consumption monitoring data and environmental parameter data in the data center operating environment, perform time-series analysis on the energy consumption monitoring data to identify energy consumption patterns, and generate graded quality parameters based on the quality requirements of the energy consumption monitoring data. The marginal analysis module is used to perform marginal utility evaluation based on the energy consumption pattern and the graded quality parameters to determine the precision configuration boundary, perform impact intensity evaluation on the precision configuration boundary to generate a carbon source intensity index, and use spatial mapping processing to convert the carbon source intensity index into an impact coverage area. The propagation analysis module is used to analyze the change trajectory of the environmental parameter data to generate a carbon propagation network, perform frequency analysis on the carbon propagation network to obtain key propagation paths, map the key propagation paths to carbon mobility indicators, and reconstruct the monitoring network based on the carbon mobility indicators to form a path optimization table. The range control module is used to assess the abnormal propagation range of the affected coverage area, determine the monitoring extension area, perform dynamic range adjustment on the monitoring extension area to generate range adjustment parameters, integrate the range adjustment parameters with the path optimization table to construct an adaptive monitoring configuration, and identify emission reduction potential areas based on the adaptive monitoring configuration. The matrix processing module is used to decompose the emission reduction potential area into spatial monitoring nodes, allocate utility weights to the spatial monitoring nodes to generate differentiated monitoring sequences, perform matrix operations based on the differentiated monitoring sequences to generate an energy consumption and carbon emission analysis matrix, perform statistical evaluation on the energy consumption and carbon emission analysis matrix to generate monitoring configuration parameters, and complete intelligent monitoring of energy consumption and carbon emission in the data center.

[0007] The beneficial effects of this invention are reflected in the following points: 1. By monitoring energy consumption data and environmental data, energy consumption patterns are identified and a three-level quality configuration system is generated, realizing differentiated allocation and precise deployment of energy consumption monitoring resources. Marginal utility assessment quantifies and matches monitoring accuracy with actual benefits, effectively improving the cost-effectiveness and operational efficiency of the monitoring system. 2. Based on the trajectory of environmental parameter data changes, a carbon propagation network is constructed and key propagation paths are identified, transforming the abstract carbon emission diffusion process into a specific network topology and propagation channels. Carbon mobility indicators can quantify the activity level and impact range of carbon emission propagation in different regions. Through propagation frequency characteristic analysis and low-frequency propagation area identification, intelligent optimization of the monitoring network layout and precise adjustment of propagation signals are achieved, solving the problem that traditional static monitoring layouts cannot adapt to dynamic changes in carbon emissions. 3. The system employs anomaly propagation range assessment and dynamic range adjustment techniques to achieve intelligent expansion of monitoring boundaries. By allocating utility weights for spatial monitoring nodes and identifying redundant areas, it solves the problems of uneven allocation of monitoring resources and duplicate monitoring. It constructs differentiated monitoring sequences and transforms redundant data into valuable analytical resources. By simplifying matrix operations and using recursive chains, it reduces computational complexity and achieves a shift from extensive monitoring management to refined intelligent monitoring. While ensuring the integrity of monitoring coverage, it optimizes system operating efficiency and the rationality of resource allocation.

[0008] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0009] The accompanying drawings illustrate specific examples of the technical solutions described in this invention and, together with the detailed embodiments, form part of the specification, serving to explain the technical solutions, principles, and effects of this invention.

[0010] Unless otherwise specified or defined, the same reference numerals in different figures represent the same or similar technical features, and different reference numerals may be used to represent the same or similar technical features.

[0011] Figure 1 This is a flowchart illustrating an intelligent monitoring method for energy consumption and carbon emissions in data centers according to the present invention.

[0012] Figure 2 This is a structural block diagram of an intelligent monitoring system for energy consumption and carbon emissions in a data center according to the present invention. Detailed Implementation

[0013] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0014] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0015] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0016] The technical solutions of the embodiments of this application will be described below.

[0017] like Figure 1 As shown, this embodiment of the invention provides a smart monitoring method for energy consumption and carbon emissions in data centers, including the following steps S110-S150: Step S110: Monitor energy consumption monitoring data and environmental parameter data in the data center operating environment, perform time-series analysis on the energy consumption monitoring data to identify energy consumption patterns, and generate graded quality parameters based on the quality requirements of the energy consumption monitoring data.

[0018] Specifically, the system monitors energy consumption and environmental parameters within the data center's operating environment. Intelligent power monitoring devices are deployed in key areas of the data center to collect real-time power consumption data from server rooms, storage areas, network equipment areas, and cooling systems. These power monitoring devices utilize high-precision power analyzers, capable of simultaneously measuring multiple energy consumption indicators such as active power, reactive power, apparent power, and power factor. The sampling frequency is set to a 10-second interval to ensure the capture of detailed characteristics of energy consumption changes. Intelligent PDU devices are installed at the server rack level to monitor the real-time power consumption and cumulative energy usage of each rack. Simultaneously, a multi-point environmental parameter monitoring network is deployed within the data center, including temperature sensors, humidity sensors, CO2 concentration sensors, and air quality monitoring equipment. Temperature sensors are distributed at air inlets, outlets, and hot aisles in the server room; humidity sensors are installed in pairs with temperature sensors; and CO2 concentration sensors are primarily deployed in areas with high personnel activity and at key nodes of the ventilation system. Environmental parameter data is collected every minute and transmitted to the data acquisition center via both wired and wireless sensor networks. Energy consumption monitoring nodes are installed at the UPS system, diesel generator set, and mains power input of the data center to comprehensively monitor the energy consumption of each link in the power supply chain.

[0019] Time-series analysis of energy consumption monitoring data is performed to identify energy consumption patterns. The collected energy consumption monitoring data is arranged according to time series, and a sliding window technique is used to segment the time-series data, with a window length of 24 hours and a step size of 1 hour, forming continuous energy consumption time-series segments. Fast Fourier Transform (FFT) analysis is used to analyze the frequency domain characteristics of the energy consumption time-series data, identifying the periodic components and main frequency components of energy consumption changes. In the time domain analysis, a trend decomposition method is used to decompose the energy consumption time-series data into three basic components: trend, periodic, and random fluctuation. Typical energy consumption patterns in the data center are identified, including four main types: basic load mode, peak business mode, maintenance operation mode, and emergency response mode. The basic load mode corresponds to the energy consumption level of the data center under minimum operating conditions; the peak business mode corresponds to the energy consumption increase during periods of concentrated user access; the maintenance operation mode corresponds to energy consumption fluctuations during system maintenance; and the emergency response mode corresponds to energy consumption surges under sudden events. Cluster analysis is used to classify similar energy consumption time-series patterns, identifying the typical characteristics and changing patterns of each type of pattern. In actual operation, it was observed that the Web service cluster exhibited a clear peak energy consumption pattern from 9 am to 6 pm on weekdays, while the storage system showed a data backup maintenance operation energy consumption pattern from 2 am to 5 am at night.

[0020] Based on the quality requirements of energy consumption monitoring data, tiered quality parameters are generated. The differences in quality requirements for energy consumption monitoring data under different application scenarios are analyzed, and the quality requirements are divided into three levels: precise monitoring, standard monitoring, and coarse monitoring. The precise monitoring level is suitable for critical business systems and high-energy-consuming equipment, requiring data acquisition accuracy within 1%, data integrity exceeding 99.9%, and time latency controlled within 10 seconds. The standard monitoring level is suitable for general business systems and auxiliary equipment, requiring data acquisition accuracy within 5%, data integrity exceeding 99%, and time latency controlled within 1 minute. The coarse monitoring level is suitable for backup equipment and non-critical systems, requiring data acquisition accuracy within 10%, data integrity exceeding 95%, and acceptable time latency within 5 minutes. Based on the characteristics of the requirements for different quality levels, corresponding sets of tiered quality parameters are generated. Quality parameters for the precise monitoring level include high sampling frequency, multiple verification mechanisms, redundant data backup, and real-time anomaly detection. Quality parameters for the standard monitoring level include medium sampling frequency, basic verification mechanisms, and periodic data backup. Quality parameters for the coarse monitoring level include low sampling frequency, simplified verification mechanisms, and batch data storage. Considering the importance and energy consumption characteristics of different monitoring objects, corresponding quality levels are assigned to various types of equipment and systems. Core servers, main storage systems, and critical network equipment are assigned precise monitoring levels; general servers and auxiliary storage systems are assigned standard monitoring levels; and backup equipment and environmental control systems are assigned coarse monitoring levels.

[0021] Step S120: Based on the energy consumption pattern and graded quality parameters, perform marginal utility evaluation to determine the precision configuration boundary, perform influence intensity evaluation on the precision configuration boundary to generate a carbon source intensity index, and use spatial mapping processing to convert the carbon source intensity index into an influence coverage area.

[0022] In some embodiments, determining the precision configuration boundary by performing marginal utility evaluation based on the energy consumption pattern and the graded quality parameters includes: identifying the peak frequency band of consumption intensity for the energy consumption pattern; generating utility correlation points by performing utility correlation between the peak frequency band of consumption intensity and the graded quality parameters; using the utility correlation points to stimulate changes in the precision demand distribution to generate a precision demand domain; and determining the precision configuration boundary through the precision demand domain.

[0023] This study identifies peak frequency bands for energy consumption intensity based on consumption patterns. Frequency domain analysis is performed on the four energy consumption patterns output from step S110 to identify the main frequency components of energy intensity changes under each pattern. Power spectral density analysis is used to extract the spectral characteristics of each consumption pattern, identifying the frequency ranges where energy is concentrated. Under the basic load pattern, energy consumption changes are mainly concentrated in the low-frequency band of 0.01-0.05Hz, reflecting the slow power consumption adjustment process of the equipment. Under the peak service pattern, energy consumption fluctuations are mainly distributed in the mid-frequency band of 0.1-0.5Hz, corresponding to rapid changes in user access volume. Under the maintenance operation pattern, energy consumption changes exhibit ultra-low frequency characteristics of 0.005-0.02Hz, reflecting the long-cycle characteristics of maintenance tasks. Under the emergency response pattern, energy consumption surges are concentrated in the high-frequency band of 0.5-2Hz, reflecting the rapid response characteristics of the system. Peak detection algorithms are used to identify energy peak points within each frequency band, determining the frequency positions corresponding to the peak consumption intensity. The frequency components within a certain bandwidth around the peak frequency are defined as the peak consumption intensity frequency band. Analyzing the bandwidth characteristics of peak frequency bands under different consumption patterns reveals that wider bandwidth indicates more complex energy consumption changes, requiring higher monitoring accuracy. When data center storage systems perform large-scale data migrations, their peak energy consumption bands exhibit wide-spectrum characteristics, necessitating high-precision monitoring to accurately capture the details of energy consumption changes.

[0024] Utility correlation points are generated by correlating the peak frequency band of consumption intensity with graded quality parameters. The frequency characteristics of the peak frequency band of consumption intensity are matched with the monitoring capabilities of the graded quality parameters. The high sampling frequency of the precise monitoring level effectively captures peak frequency band variations in the high-frequency band, the standard monitoring level is suitable for monitoring the mid-frequency peak frequency band, and the coarse monitoring level only meets the basic monitoring requirements of the low-frequency peak frequency band. Frequency band matching analysis determines the monitoring utility of each quality level for different peak frequency bands. The monitoring utility reaches its maximum when the sampling frequency of the monitoring level is higher than twice the upper limit frequency of the peak frequency band. When the sampling frequency is lower than the upper limit frequency of the peak frequency band, the monitoring utility drops sharply, and spectral aliasing may occur. In the utility correlation analysis, the peak frequency band of the database server's peak business mode is 0.2-0.4Hz. Using the precise monitoring level (sampling frequency 1Hz), 90% monitoring utility can be obtained, while the coarse monitoring level (sampling frequency 0.1Hz) only obtains 30% monitoring utility. Key inflection points in utility correlations are identified through utility function curve analysis. These inflection points correspond to configuration states where significant changes in monitored utility occur. These inflection points are marked as utility correlation points, each containing three elements: peak frequency band information, quality level information, and utility value.

[0025] For example, the step of using the utility correlation points to stimulate changes in the accuracy demand distribution and generate an accuracy demand domain includes: analyzing the accuracy configuration status based on the utility correlation points to generate a configuration margin table; stimulating changes in the accuracy demand distribution based on the configuration margin table to form a density optimization curve; using the density optimization curve to identify configuration paths and generate advantageous configuration paths; and performing spatial mapping on the advantageous configuration paths to generate an accuracy demand domain.

[0026] A configuration margin table is generated based on the accuracy configuration status analysis of utility correlation points. The difference between the current accuracy configuration and the optimal accuracy configuration is extracted from the utility correlation points to quantify the configuration margin of each monitoring point. A positive margin is generated when the current accuracy configuration is lower than the optimal configuration, indicating room for accuracy improvement; a negative margin is generated when the current configuration is higher than the optimal configuration, indicating configuration redundancy. The accuracy configuration margin values ​​of each device and area are measured; a larger margin indicates higher potential for configuration optimization. The current configuration parameters of each monitoring device in the data center are checked one by one using a configuration status analysis algorithm, including key indicators such as sampling frequency, data accuracy, verification method, and storage format. The check results are compared with the optimal configuration parameters determined by the utility correlation points to identify specific mismatches and the degree of deviation. In the video processing server cluster, the current standard monitoring level is used, but the optimal configuration is precise monitoring, resulting in a positive configuration margin of +0.3, indicating a need to improve monitoring accuracy to better capture GPU load changes. In the backup storage system, the current standard monitoring level is used, but the optimal configuration is coarse monitoring, resulting in a negative configuration margin of -0.2, indicating configuration overcapacity that could reduce monitoring costs. The configuration margins of each monitoring point are categorized and organized according to equipment type, spatial location, and temporal characteristics to form a structured configuration margin table.

[0027] A density optimization curve is generated based on the changes in accuracy requirement distribution triggered by the configuration margin table. A kernel density estimation method is used to fit the probability density distribution of the configuration margin, identifying concentrated and sparse regions of configuration requirement. Hotspots of accuracy requirement are determined by the peak positions of the density distribution curve; higher peaks indicate more urgent configuration optimization needs in those regions. The gradient of the density curve is analyzed; regions with larger gradients indicate drastic changes in accuracy requirement, requiring close attention and rapid response. The density distribution is visualized in three dimensions according to spatial coordinates, forming a three-dimensional distribution image of accuracy requirement. In areas with higher temperatures in the data center, due to increased equipment power consumption fluctuations and cooling demands, the accuracy requirement density curve exhibits a distinct peak characteristic. In the core switch area of ​​the data center, due to frequent changes in network traffic, the accuracy requirement density curve shows a multi-peak distribution, reflecting complex monitoring demand patterns. A density gradient analysis method is introduced to identify the boundary locations and transition regions where accuracy requirement changes drastically. The fitted density distribution curve is used as the density optimization curve; the shape and parameters of the curve reflect the spatial distribution pattern and optimization direction of accuracy requirement.

[0028] A density optimization curve is used to identify optimal configuration paths. The optimal adjustment path for precision configuration is identified along the gradient direction of the density optimization curve; the gradient upward direction corresponds to adjustment paths with increased precision requirements, and the gradient downward direction corresponds to adjustment paths with decreased precision requirements. A path search algorithm is used to find the shortest path connecting the current configuration state and the target configuration state on the density optimization curve. The A* search algorithm is used for path planning, and its heuristic function combines multiple factors such as path length, adjustment cost, and implementation difficulty. Constraints on configuration adjustment are considered, including hardware capability limitations, cost control requirements, implementation complexity constraints, and time window limitations. The impact of various constraints on path selection is analyzed; hard constraints must be strictly met, while soft constraints can be relaxed to a certain extent. During path selection, priority is given to path options with moderate configuration adjustment ranges, low implementation difficulty, and controllable risks. The path evaluation index system includes adjustment effect indicators, cost-effectiveness indicators, implementation feasibility indicators, and risk assessment indicators. The dominant path among multiple candidate configuration paths is identified; the dominant path should possess the comprehensive characteristics of significant adjustment effect, low resource consumption, controllable implementation risk, and low maintenance cost.

[0029] For advantageous configuration paths, spatial mapping is used to generate precision requirement domains. A coordinate transformation algorithm is employed to convert abstract configuration parameters into specific device locations and spatial coordinates. Based on the device locations and influence ranges involved in the configuration path, corresponding spatial areas are marked on the data center floor plan. Considering the physical connections and functional dependencies between devices, the spatial range of the path's influence is expanded. Multiple dispersed spatial areas involved in the same advantageous configuration path are connected and merged, and a spatial clustering algorithm is used to form a continuous precision requirement domain. Morphological operations are applied to optimize the shape of irregular spatial areas, generating regular areas that are easy to manage and implement. The boundary of the precision requirement domain is jointly determined by the influence range of the configuration path and the device distribution density; all monitoring points within the domain need to be adjusted for precision configuration according to this path. In the storage array area of ​​the data center, due to the dense distribution of disk devices, its precision requirement domain exhibits a long, narrow shape, extending along the arrangement direction of the storage racks. The geometric characteristics of the precision requirement domain are analyzed, including parameters such as area, boundary perimeter, shape complexity, and spatial connectivity.

[0030] Precision configuration boundaries are determined through precision requirement domains. Boundary information and geometric features of the precision requirement domains are used to delineate the boundaries of areas within the data center with different precision configuration requirements. The outer boundary of the precision requirement domain is used as the initial candidate line for precision configuration boundaries. Boundary optimization algorithms are used to geometrically adjust and topologically optimize the candidate lines. A boundary smoothing algorithm is employed to eliminate jagged irregularities in the boundary lines, generating smooth and continuous configuration boundaries. Consideration is given to the continuity and consistency requirements of precision configuration, avoiding configuration inconsistencies and management complexity caused by excessive differences in precision levels between adjacent areas. The physical layout characteristics of the data center are fully considered during boundary delineation. A transition buffer zone is set between the network device area and the server area, with medium-precision configuration used within the buffer zone to ensure a smooth transition in configuration levels. In the cooling system area of ​​the data center, where environmental parameters change relatively slowly, the configuration boundary can be set relatively loosely, and a coarse monitoring level is sufficient to meet the requirements. Boundary adjustments are only triggered when business requirements change significantly or equipment configurations are extensively updated. Considering the practicality of boundary management, complex irregular boundaries are simplified into easily identifiable and manageable regular geometric shapes. The determined precision configuration boundary is overlaid with the data center equipment distribution map, network topology map, and management area map to generate a complete precision configuration boundary management map.

[0031] A carbon source intensity index is generated by assessing the impact of the precision configuration boundary. The influence of energy consumption characteristics at each monitoring point within the precision configuration boundary on carbon emissions is analyzed, quantifying the carbon source intensity of different equipment and regions. An energy consumption-carbon emission conversion coefficient is used to convert the electricity consumption of each monitoring point into corresponding carbon emissions. Considering the grid carbon emission factor of the data center's location, the conversion coefficient increases when the proportion of local thermal power is high and decreases when the proportion of renewable energy is high. The variation pattern of the grid carbon emission factor at different times is analyzed; the grid carbon emission factor is typically higher during peak electricity consumption at night and relatively lower during the day when solar power generation is sufficient. A carbon source intensity weight is assigned to each monitoring point based on the equipment's operating characteristics and energy consumption intensity. High-power GPU computing clusters and large storage arrays are assigned higher carbon source intensity weights, while low-power network switching equipment and environmental monitoring systems are assigned lower weights. The instantaneous carbon source intensity value for each monitoring point is generated by multiplying the carbon source intensity weight by the actual energy consumption data. The instantaneous carbon source intensity value is then averaged over time to obtain a stable carbon source intensity index. In the AI ​​training area of ​​the data center, the carbon source intensity index is usually 3-5 times that of ordinary server areas because the GPU clusters run at high load for a long time.

[0032] In some embodiments, the step of converting the carbon source intensity index into an impact coverage area using spatial mapping processing includes: identifying differences in mapping density based on the carbon source intensity index to generate dense and sparse mapping areas; using the sparse mapping area to perform impact range inference processing to form an extended inference area; spatially fusing the extended inference area with the dense mapping area to generate a complete coverage segment; and spatially optimizing the complete coverage segment to construct an impact coverage area.

[0033] Based on the carbon source intensity index, spatial mapping is used to identify density differences, resulting in dense and sparse mapping regions. The carbon source intensity index is gridded according to spatial coordinates, with each grid cell corresponding to a 5m × 5m physical area within the data center. Spatial interpolation methods are used to map the discrete carbon source intensity index to a continuous grid space, employing an inverse distance weighted interpolation algorithm to ensure the smoothness of the mapping results. The density distribution of the carbon source intensity index within each grid cell is analyzed; high-density grids indicate a concentration of multiple high-intensity carbon sources. When the carbon source intensity index density within a grid exceeds 1.5 times the average density, the grid is marked as a dense mapping region. When the carbon source intensity index density within a grid is less than 0.5 times the average density, the grid is marked as a sparse mapping region. Dense mapping regions typically correspond to areas with concentrated high-power devices within the data center, such as GPU computing clusters and large storage arrays. Sparse mapping regions typically correspond to areas with lower device density, such as aisles, rest areas, and equipment maintenance spaces. In the AI ​​training area of ​​the data center, due to the high-density deployment of GPU servers, almost the entire area is marked as a dense mapping region, with a carbon source intensity index density reaching 4-6 times the average. Adjacent dense mapping grids are merged into continuous dense mapping regions through connected component analysis, and adjacent sparse mapping grids are merged into continuous sparse mapping regions.

[0034] A sparse mapping region is used to infer the influence range and form an extended inference region. The spatial adjacency relationship between sparse and dense mapping regions is analyzed to identify potential connecting channels where carbon source diffusion influences may occur. Considering airflow patterns within the data center, sparse mapping regions along cold and hot aisles are more susceptible to diffusion from adjacent dense mapping regions. Diffusion propagation simulation is used to predict the degree of carbon source intensity propagation from dense to sparse mapping regions; the propagation intensity decreases exponentially with distance. When the distance between a sparse and dense mapping region and the dense region is less than 10 meters and the region is downwind, it is marked as a high-influence expansion candidate region. The boundary range of the extended inference region is determined through influence propagation path analysis, which is influenced by factors such as airflow velocity, temperature gradient, and obstacle distribution. In the server room of the data center, the sparse mapping region near the air conditioning vents has a relatively large extended inference region due to airflow disturbance, extending up to 15 meters from the dense region. In the battery storage area of ​​the data center, although the equipment density is low and it is marked as a sparse mapping region, the heat diffusion during battery charging and discharging extends the area within 5 meters to the extended inference region. The inferred influence extension area is marked as the extended inference region, and the source of influence and expected influence intensity of the inference region are recorded.

[0035] A complete coverage segment is generated by spatially fusing extended inference regions and dense mapping regions. Spatial union operation expands the boundary of the dense mapping region outwards to encompass the relevant extended inference region. During the fusion process, the weight differences of different influencing regions are considered; the influence weight of the dense mapping region is set to 1.0, and the influence weight of the extended inference region is set to 0.3-0.7 based on distance attenuation characteristics. A weighted fusion method is used to handle the influence intensity of overlapping regions; the comprehensive influence intensity of the overlapping region is equal to the weighted sum of the weights of each influencing source. Boundary smoothing is applied to the fused coverage area to eliminate jagged, irregular shapes. The connectivity characteristics of the complete coverage segment are analyzed to identify independent coverage segments and interconnected coverage segment groups. In the network core area of ​​the data center, due to the relatively dispersed nature of switches and routers, multiple independent complete coverage segments are formed after fusion, each covering an area of ​​approximately 50-100 square meters. Coverage segments with similar shapes and close proximity are further merged to generate a larger-scale complete coverage segment. Considering the functional correlation between coverage segments, coverage segments serving the same business system are prioritized for merging.

[0036] Spatial optimization is performed to construct the impact coverage area for complete coverage segments. Geometric optimization is applied to the generated complete coverage segments to improve the regularity and usability of the coverage area. A convex hull algorithm is used to regularize the shape of irregular coverage segments, generating regular-shaped coverage areas that are easy to manage and control. Considering the actual situation of the channel layout and equipment distribution within the data center, the coverage area boundaries are adjusted to adapt to physical constraints. The coverage area boundaries are aligned with the main channels, equipment boundaries, and security areas within the computer room to ensure the actual operability of the coverage area. Coverage segments that are too small are merged or deleted to avoid generating too many unmanageable tiny coverage areas. A minimum area threshold of 25 square meters is set for the coverage area; segments smaller than this threshold are merged with adjacent larger segments. In the UPS power supply area of ​​the data center, due to the large size and relatively fixed distribution of the equipment, the optimized impact coverage area presents a regular rectangular shape, facilitating identification and management by maintenance personnel. The optimized coverage segments are classified according to their impact intensity, with different identification and management strategies applied to strong-impact, medium-impact, and weak-impact coverage areas.

[0037] Step S130: Analyze the change trajectory of environmental parameter data to generate a carbon propagation network, perform frequency analysis on the carbon propagation network to obtain key propagation paths, map the key propagation paths to carbon mobility indicators, and reconstruct the monitoring network based on the carbon mobility indicators to form a path optimization table.

[0038] Specifically, a carbon propagation network is generated by analyzing the changing trajectories of environmental parameter data. Environmental parameter data such as temperature, humidity, CO2 concentration, and air quality collected in step S110 are arranged in time series, and the changing trajectories of each parameter at different spatial locations are analyzed. Gradient analysis of the time series data identifies the direction and rate of change of environmental parameters; regions with large gradients indicate drastic changes in environmental parameters, corresponding to active carbon emission propagation areas. Correlation analysis is used to identify the degree of correlation between environmental parameter changes at different monitoring points; when the CO2 concentration changes at two monitoring points show a high correlation, it indicates a carbon emission propagation link. Monitoring points with strong correlations are connected to form the basic network structure for carbon propagation. In the server room of the data center, the CO2 concentration changes in the GPU cluster area and the CO2 concentration changes at the downstream hot aisle outlet show a significant time-lag correlation, indicating that carbon emissions propagate along the airflow direction. The physical path of carbon propagation is determined through airflow analysis, and the main carbon propagation directions are identified by combining the supply and return air patterns of the ventilation system. The propagation paths are weighted according to propagation intensity, which is determined by both the correlation coefficient and the time-lag characteristics. The importance of nodes in a network is determined by both connectivity and propagation influence. Important nodes typically correspond to key sources or accumulation points of carbon emissions. In data center cooling rooms, due to the convergence of waste heat from multiple heat aisles, this area becomes a significant accumulation node in the network.

[0039] In some embodiments, the step of performing frequency analysis on the carbon propagation network to obtain key propagation paths includes: obtaining the propagation frequency characteristics of the carbon propagation network; identifying low-frequency propagation regions based on the propagation frequency characteristics to generate low-frequency region parameters; passively adjusting the propagation signal using the low-frequency region parameters to generate an adjusted propagation signal; and identifying key propagation paths using the adjusted propagation signal.

[0040] This study aims to obtain the propagation frequency characteristics of the carbon propagation network. Frequency domain analysis is performed on the activation modes of each propagation path in the network to identify the frequency distribution characteristics of propagation activities. The Fast Fourier Transform (FFT) method is used to convert the time-domain activation signals of the propagation paths into frequency-domain signals, extracting the spectral characteristics of each path. The distribution range of propagation frequencies is analyzed. Carbon propagation frequencies within data centers are typically distributed in the low-frequency band of 0.001Hz-0.1Hz, corresponding to propagation cycles ranging from a few minutes to several hours. On the main propagation path in the server room, the propagation frequencies are mainly concentrated in the 0.01Hz-0.05Hz band, corresponding to propagation cycles of 20 to 100 seconds, reflecting carbon emission fluctuations caused by changes in server load. The dominant frequency components of different propagation paths are identified, with the dominant frequency corresponding to the most active propagation mode of that path. Power spectral density analysis is used to determine the energy distribution of each frequency component; the frequency bands with concentrated energy represent the main characteristic frequencies of propagation activities. The temporal stability of the propagation frequencies is analyzed. Stable frequency components reflect the inherent propagation characteristics of the system, while unstable frequency components may be related to external interference or abnormal events.

[0041] Low-frequency propagation regions are identified and low-frequency region parameters are generated based on propagation frequency characteristics. Propagation paths and nodes with propagation frequencies below 0.005 Hz in the carbon propagation network are scanned; these low-frequency regions correspond to areas with relatively slow and stable carbon propagation activity. The spatial distribution characteristics of low-frequency propagation regions are analyzed to identify their clustering and dispersion patterns. Low-frequency propagation regions typically correspond to data center areas with stable equipment operation and slow environmental changes, such as backup equipment areas, battery storage areas, and equipment maintenance areas. The propagation delay characteristics of low-frequency propagation regions are measured; the propagation delay in low-frequency regions is usually relatively long, meaning changes in carbon emission concentration require a considerable amount of time to propagate to downstream monitoring points. In the archive storage area of ​​a data center, due to the low usage frequency and stable power consumption of tape library equipment, this area is identified as a typical low-frequency propagation region, with a propagation frequency of only 0.001 Hz. Low-frequency region parameters are generated, including four core indicators: average propagation frequency, propagation delay time, region coverage area, and node connectivity. The average propagation frequency reflects the overall frequency level of propagation activity within the region; the propagation delay time reflects the time lag characteristics of the propagated signal; the region coverage area reflects the spatial extent of the low-frequency region; and the node connectivity reflects the degree of connection between the region and other regions.

[0042] Passive adjustment of the propagation signal is achieved using low-frequency region parameters. Based on the characteristics of the low-frequency region parameters, the propagation signal is passively adjusted. Passive adjustment means that the physical characteristics of the propagation path are not actively changed, but the quality of the propagation signal is optimized through signal processing techniques. Signal amplification technology is used to enhance the weak propagation signal in the low-frequency region. The amplification factor is determined according to the propagation frequency characteristics of the region, with higher amplification factors used for lower frequency regions. Filtering technology is used to eliminate high-frequency noise components in the propagation signal while retaining the effective signal characteristics of low-frequency propagation. In the isolated areas of the data center server room, due to the relatively enclosed environment and less interference, the propagation signal quality in this area is high, requiring only minor signal adjustment. Signal delay compensation technology is used to correct the propagation delay effect in the low-frequency propagation region, ensuring that the propagation signals in different areas remain synchronized in time. Adaptive threshold adjustment technology is used to dynamically adjust the detection threshold of the propagation signal, optimizing the threshold adjustment based on the propagation intensity and noise level of the region. The adjusted propagation signal is compared and analyzed with the original propagation signal to evaluate the effect and degree of improvement of the signal adjustment. In the ventilation duct area of ​​the data center, due to the greater noise generated by airflow disturbances, the propagation signal in this area requires stronger filtering and noise reduction processing. The generated modulated propagation signal has a higher signal-to-noise ratio and more stable propagation characteristics, and can more accurately reflect the true state of carbon propagation.

[0043] Identifying critical propagation paths by modulating propagation signals. The importance and efficiency of carbon propagation networks are re-analyzed using the modulated propagation signals. Strong signal paths within the modulated propagation signals are identified through signal strength analysis; these paths correspond to propagation channels with high propagation activity and good signal quality. Signal correlation analysis is used to identify the degree of association between modulated propagation signals; highly correlated signal paths indicate close propagation connections. A path tracing algorithm is used to identify complete propagation paths along the propagation direction of the modulated propagation signals, starting from the propagation source and extending along the signal strength gradient to the propagation endpoint. The propagation distance and speed characteristics of the modulated propagation signals are analyzed; paths with longer propagation distances and faster propagation speeds have higher propagation efficiency. In the core server room area of ​​a data center, the propagation path from the GPU computing cluster to the precision air conditioning return air vent exhibits strong propagation signal characteristics after signal modulation and is identified as a primary critical propagation path. Multi-path comparative analysis is used to determine the importance ranking of critical propagation paths, based on comprehensive indicators including modulated signal strength, propagation range, propagation frequency, and propagation stability. Key nodes within the critical propagation paths are identified; these nodes are typically located at the intersection of multiple paths or at propagation inflection points.

[0044] Key propagation paths are mapped to carbon mobility indicators. A characterization system for carbon mobility indicators is constructed based on characteristic parameters such as activation frequency, propagation intensity, and path length of key propagation paths. The carbon mobility indicator reflects the activity and propagation efficiency of carbon emissions along specific paths. A weighted synthesis method is used to fuse multiple path characteristic parameters into a single mobility indicator value, with weights determined according to the importance of each parameter to carbon propagation. The weight of activation frequency is set to 0.4, reflecting the frequency of path use; the weight of propagation intensity is set to 0.3, reflecting the effectiveness of path propagation; the weight of path length is set to 0.2, reflecting the distance characteristics of propagation; and the weights of other characteristic parameters are set to 0.1. Normalization is used to unify the numerical range of the carbon mobility indicator to 0-1, facilitating comparative analysis between different paths. In the network equipment area of ​​data centers, due to the relatively stable and uniform power consumption of switches and routers, the carbon mobility indicator in this area exhibits a moderately uniform distribution. The spatial distribution pattern of the carbon mobility indicator is analyzed, identifying three levels: high mobility areas, medium mobility areas, and low mobility areas. High mobility areas typically correspond to areas with active carbon emissions and dense propagation paths, requiring focused monitoring and control. The carbon mobility index is mapped to the location of the monitoring point, and a corresponding mobility index value is assigned to each monitoring point.

[0045] A path optimization table is generated by reconstructing the monitoring network based on the carbon mobility index. The numerical distribution of the carbon mobility index is used to reassess the rationality of the existing monitoring network layout and its monitoring effectiveness. The matching degree between monitoring point density and the carbon mobility index is analyzed; higher density monitoring points should be configured in high-mobility areas, while the density can be appropriately reduced in low-mobility areas. The spatial configuration of monitoring points is adjusted through a monitoring network optimization algorithm, with the optimization objective being to maximize monitoring coverage under monitoring cost constraints. Redundant monitoring points in the monitoring network are identified; when multiple monitoring points are located in low-mobility areas and their monitoring ranges overlap, merging or removing some monitoring points can be considered. In the office areas of data centers, due to less personnel activity and lower equipment density, the carbon mobility index is generally lower, and the number of monitoring points or the monitoring frequency can be reduced. Blank areas in the monitoring network are identified; when high-mobility areas lack sufficient monitoring coverage, it is necessary to add monitoring points or adjust the positions of existing monitoring points. A new monitoring network topology is generated by reconfiguring monitoring points; the new network should be able to better capture the dynamic characteristics of carbon propagation. The adjustment schemes of the monitoring network are compiled into a path optimization table, which includes fields such as monitoring point identifier, original location, target location, adjustment type, mobility index, and implementation priority. The path optimization table is sorted according to implementation priority. Higher-priority adjustments are implemented first, while lower-priority adjustments can be postponed depending on resource availability. In the UPS power room of the data center, due to stable equipment operation and a closed environment, the monitoring points in this area have a lower priority for adjustments and can be implemented uniformly during system maintenance.

[0046] Step S140: Assess the abnormal propagation range of the affected coverage area to determine the monitoring extension area, perform dynamic range adjustment on the monitoring extension area to generate range adjustment parameters, integrate the range adjustment parameters with the path optimization table to construct an adaptive monitoring configuration, and identify emission reduction potential areas based on the adaptive monitoring configuration.

[0047] In some embodiments, assessing the abnormal propagation range of the affected coverage area to determine the monitoring extension area includes: identifying low-anomaly regions by performing propagation range anomaly measurement based on the affected coverage area; using the low-anomaly regions as a stability reference to form a reference quality distribution; using the reference quality distribution to derive range values ​​to generate a derived range matrix; and performing spatial mapping on the derived range matrix to determine the monitoring extension area.

[0048] Based on the impact coverage area, anomaly measurement of propagation range is used to identify low-anomaly areas. The degree of anomaly in carbon emission propagation data from each monitoring point within the impact coverage area output in step S120 is assessed, quantifying the degree of deviation from the normal propagation pattern in each area. Statistical analysis methods are used to determine the normal range benchmark for carbon emission propagation. The benchmark range is determined by the statistical characteristics of historical data, including statistical parameters such as mean, standard deviation, and coefficient of variation. The degree of propagation anomaly in each area is measured using an anomaly measurement formula, where the degree of anomaly equals the deviation of the actual propagation range from the normal benchmark range divided by the standard deviation of the benchmark range. When the degree of anomaly is less than one standard deviation, the area is marked as a low-anomaly area, indicating that the carbon emission propagation behavior in that area is relatively stable and predictable. In the UPS power supply area of ​​the data center, due to the stable operation of the UPS equipment and the enclosed environment, the carbon emission propagation range in this area changes very little, with an anomaly of only 0.3 standard deviations, and is identified as a typical low-anomaly area. The spatial distribution characteristics of low-anomaly areas are analyzed to identify their clustering patterns and distribution regularities. Low-anomaly areas typically correspond to data center areas with stable equipment operation, consistent environmental conditions, and minimal human interference. Connectivity analysis merges adjacent low-anomaly regions into a continuous, stable propagation region, resulting in a more statistically representative area. In the fiber optic cabling area of ​​a data center, because fiber optic equipment generates virtually no heat and has no moving mechanical parts, this area maintains a low-anomaly state for over 30 days, becoming the most stable low-anomaly region.

[0049] A reference quality distribution is formed by using low-anomaly regions as stability references. Identified low-anomaly regions are used as benchmarks for stable propagation, and the carbon emission propagation quality characteristics and distribution patterns in these regions are analyzed. Propagation quality parameters are extracted from low-anomaly regions, including key indicators such as propagation consistency, propagation predictability, propagation stability, and propagation efficiency. Propagation consistency reflects the similarity of propagation behavior at different times; propagation predictability reflects the regularity of propagation behavior; propagation stability reflects the fluctuation range of propagation parameters; and propagation efficiency reflects the energy utilization efficiency of the propagation process. The distribution characteristics of propagation quality parameters in low-anomaly regions are fitted using probability density estimation methods to generate the probability density function of the reference quality distribution. The reference quality distribution reflects the normal distribution range and pattern of quality parameters under stable propagation conditions. In the monitoring room area of ​​the data center, due to the low power consumption and continuous stable operation of the monitoring equipment, the propagation quality parameters in this area exhibit a highly concentrated normal distribution with a small variance, making it an ideal source of reference quality distribution. The statistical characteristics of the reference quality distribution are analyzed, including parameters such as the mean, variance, skewness, and kurtosis. These statistical characteristics are used for subsequent quality assessment and range derivation. By fusing multi-regional reference quality distributions, a more representative comprehensive reference quality distribution is generated.

[0050] A range value derivation matrix is ​​generated using a reference quality distribution. Based on the statistical characteristics and distribution parameters of the reference quality distribution, reasonable propagation range values ​​for other areas are derived. The propagation quality parameters of the area to be evaluated are compared with the reference quality distribution using a quality similarity matching method to determine the quality similarity level. When the quality parameters of the area to be evaluated fall within one standard deviation of the reference quality distribution, the area can adopt a propagation range configuration similar to that of low-anomaly areas. Reasonable propagation range values ​​for the area to be evaluated are derived based on quality similarity using linear interpolation and nonlinear fitting methods. Areas with higher quality similarity have derived range values ​​closer to the reference area's range value; areas with lower quality similarity require appropriate adjustments to their range values. In the standby server area of ​​the data center, since the equipment is in standby mode and the environmental conditions are similar to the reference area, a reasonable propagation range of 12-18 meters is derived from quality distribution matching, which is close to the reference area's baseline value of 15 meters. The derived range values ​​are organized into a matrix according to spatial coordinates and equipment type, with the row coordinates corresponding to spatial location and the column coordinates corresponding to range parameter types. The derived range matrix includes multiple parameter columns such as minimum propagation range, maximum propagation range, average propagation range, and range confidence level. Numerical analysis of the matrix elements identifies the spatial distribution patterns and gradients of the range values; regions with larger gradients indicate drastic changes in the propagation range, requiring more refined monitoring configurations.

[0051] Spatial mapping is performed on the derived range matrix to determine the monitoring extension area. The range values ​​in the derived range matrix are mapped from the parameter space to the physical space of the data center, generating specific monitoring extension area boundaries. A spatial interpolation algorithm is used to extend the discrete range values ​​into a continuous spatial range field, with radial basis function interpolation ensuring spatial continuity. The extended monitoring boundary for each monitoring point is determined based on the maximum propagation range value in the derived range matrix; the extended boundary should cover the maximum possible propagation influence range of that point. Using buffer analysis, circular or elliptical monitoring extension areas are generated with each monitoring point as the center and the derived propagation range as the radius. Considering the influence of obstacles and boundary constraints within the data center, the theoretical extension area is geometrically corrected and its boundaries are trimmed. Near the walls of the data center server room, the monitoring extension area is limited by physical boundaries; the extension range needs to be adjusted according to the wall location to avoid extending into invalid areas outside the server room. Spatial union operations are used to merge the extension areas of multiple monitoring points to generate a fully covered monitoring extension area.

[0052] Dynamic range adjustment parameters are generated for the monitoring extension area. For a given monitoring extension area, the dynamic range adjustment needs and strategies are analyzed. Time series analysis is used to study the dynamic change patterns of carbon emission concentration within the monitoring extension area, identifying peak and plateau periods. During peak periods, the monitoring extension area requires higher monitoring density and faster response times; during plateau periods, monitoring intensity can be reduced to conserve resources. An adaptive threshold adjustment method is used to dynamically adjust the boundary range of the monitoring extension area, expanding the monitoring range when carbon emission activity increases and shrinking it when activity decreases. Key factors affecting range adjustment are analyzed, including equipment load changes, ambient temperature fluctuations, ventilation system operation status, and personnel activity intensity. In the storage array area of ​​the data center, due to a significant increase in disk read / write load during nighttime backups, the monitoring extension area needs to be expanded by 20% at night to cover the increased carbon emission propagation. The generated range adjustment parameters include four core parameters: basic monitoring radius, peak expansion coefficient, adjustment response time, and adjustment frequency limit. The basic monitoring radius determines the minimum coverage area of ​​the extended zone, the peak expansion coefficient determines the range magnification factor during peak periods, the adjustment response time determines the execution speed of range adjustment, and the adjustment frequency limit avoids excessively frequent range adjustments.

[0053] An adaptive monitoring configuration is constructed by integrating range adjustment parameters with a path optimization table. Spatial matching analysis identifies the correspondence between monitoring points and monitoring extension areas in the path optimization table, determining which monitoring points require configuration adjustments based on the range adjustment parameters. The dynamic adjustment requirements in the range adjustment parameters are translated into specific monitoring point configuration parameters, including the monitoring point's activation status, monitoring frequency, data transmission mode, and alarm threshold. When the monitoring extension area shrinks, edge monitoring points can switch to low-power monitoring mode; when the area expands, dormant monitoring points need to be activated and switched to high-frequency monitoring mode. In the network core area of ​​the data center, where the power consumption of switching equipment is relatively stable, the adaptive monitoring configuration adopts a hybrid mode of fixed basic configuration plus dynamic adjustment. The basic configuration ensures basic monitoring needs, while dynamic adjustment responds to sudden load changes. Configuration conflicts between multiple monitoring extension areas are handled through configuration priority management; when configuration requirements of adjacent extension areas conflict, the requirements of areas with higher importance are prioritized. A parameter table for the adaptive monitoring configuration is constructed, containing fields such as monitoring point identifier, basic configuration parameters, dynamic adjustment parameters, adaptation rules, and execution conditions.

[0054] Identifying emission reduction potential areas based on adaptive monitoring configuration. High emission density areas are identified through carbon emission density analysis; these areas typically correspond to target areas with the greatest emission reduction potential. Efficiency characteristics of carbon emissions are analyzed to identify inefficient areas that generate excessive carbon emissions per unit of power consumption. Significant emission reductions can be achieved in these inefficient areas through equipment upgrades, configuration optimization, or operational mode adjustments. In older server areas of data centers, due to low equipment energy efficiency and inefficient cooling systems, these areas are identified as high-potential emission reduction areas; equipment upgrades can reduce carbon emission intensity by 30%. Time pattern analysis identifies peak-valley differences in carbon emissions; areas with significant peak-valley differences indicate emission reduction potential through load optimization and time scheduling. The relationship between equipment utilization and carbon emission efficiency is analyzed to identify wasteful areas with low equipment utilization but high carbon emission intensity. Spatial cluster analysis categorizes areas with similar emission reduction characteristics, forming a classification system for emission reduction potential areas. The identified emission reduction potential areas are prioritized according to emission reduction effectiveness and implementation difficulty, generating implementation roadmaps for these areas.

[0055] Step S150: Decompose the emission reduction potential area into spatial monitoring nodes, assign utility weights to the spatial monitoring nodes to generate differentiated monitoring sequences, perform matrix operations based on the differentiated monitoring sequences to generate an energy consumption and carbon emission analysis matrix, perform statistical evaluation on the energy consumption and carbon emission analysis matrix to generate monitoring configuration parameters, and complete the intelligent monitoring of data center energy consumption and carbon emission.

[0056] Specifically, the emission reduction potential area is decomposed into spatial monitoring nodes. The emission reduction potential area output from step S140 is decomposed into a grid according to spatial resolution, and each grid cell is transformed into a spatial monitoring node. Based on the physical layout characteristics of the data center, the monitoring nodes are differentiated according to equipment density and importance: the node spacing is set to 3 meters in high-density equipment areas, 5 meters in medium-density areas, and 8 meters in low-density areas. In the GPU computing cluster area of ​​the data center, due to the high power consumption and high carbon emission intensity of the equipment, this area is decomposed into a dense network of monitoring nodes, with a total of 24 monitoring nodes covering the entire cluster area. The functional characteristics and monitoring responsibilities of each spatial monitoring node are analyzed. Core monitoring nodes are responsible for direct monitoring of key equipment, auxiliary monitoring nodes are responsible for supplementary collection of environmental parameters, and boundary monitoring nodes are responsible for boundary detection of the propagation range. Based on the spatial location and equipment association of the monitoring nodes, a unique spatial identifier and functional code are assigned to each node. In the storage array area of ​​the data center, due to the regular arrangement of disk devices and similar monitoring needs, the monitoring nodes in this area adopt a regular grid layout for easy unified management and maintenance. The spatial monitoring nodes are grouped according to the type of monitoring object. The server monitoring node group, storage monitoring node group, network device monitoring node group, and environmental monitoring node group each undertake different monitoring tasks.

[0057] In some embodiments, the step of assigning utility weights to the spatial monitoring nodes to generate differentiated monitoring sequences includes: assigning utility weights to the spatial monitoring nodes to identify weight redundancy regions; determining adjustable margins based on the weight redundancy regions to form a margin allocation table; using the margin allocation table to optimize the monitoring path to form an optimized transmission path; and performing sequence calibration on the optimized transmission path to construct differentiated monitoring sequences.

[0058] This study identifies redundant weight regions by allocating utility weights to spatial monitoring nodes. The distribution of utility weights for each spatial monitoring node is analyzed, and regions with uneven weight allocation are identified through weight density analysis. When multiple monitoring nodes within a region have significantly higher weight values ​​than the average and their monitoring functions overlap, this region is marked as a redundant weight region. Spatial clustering analysis is used to identify the boundaries and extent of these redundant weight regions. The clustering algorithm groups monitoring nodes based on weight similarity and spatial proximity. A redundancy quantification method is used to measure the degree of weight redundancy, which is equal to the ratio of the sum of weights within the region to the theoretically optimal weight allocation. In the database server area of ​​the data center, due to the high importance level of multiple servers, the six monitoring nodes in this area are assigned high weights, with a total weight value of 4.2, while the theoretically optimal allocation only requires 2.8, resulting in significant weight redundancy. The causes of redundant weight regions are analyzed, including overestimation of importance, overlapping monitoring ranges, and improper resource allocation. Spatial analysis of redundant regions determines the scope and intensity of the redundancy impact; regions with high redundancy intensity have greater potential for weight adjustment. The identified redundant weight regions are classified according to their degree of redundancy, with high-redundancy, medium-redundancy, and slightly redundant regions corresponding to different adjustment strategies. In the backup server area of ​​the data center, although the weight of individual nodes is not high, the total weight of the area exceeds the limit due to the excessive node density, and it is identified as a medium-redundancy area.

[0059] For example, the step of determining the available reassignable surplus and forming a surplus allocation table based on the weighted redundancy region includes: assessing the surplus allocation value based on the weighted redundancy region to determine the allocation intensity, wherein the surplus allocation value includes the resource importance level, reassignability, and optimization level of surrounding nodes in the redundancy region; reallocating the surplus weights of surrounding monitoring nodes according to the allocation intensity to form a surplus weight allocation; formulating a resource allocation execution sequence using the surplus weight allocation; and forming a surplus allocation table through the resource allocation execution sequence.

[0060] The allocation intensity is determined based on the value of surplus allocation in weighted redundant areas. The allocation value of identified weighted redundant areas is assessed, comprehensively considering three key factors: resource importance level, allocateability, and the degree of optimization of surrounding nodes. Resource importance level reflects the importance of monitored resources within the redundant area to the overall monitoring effect; areas with higher importance levels require more caution when allocating resources. Allocationability reflects the flexibility of redundancy weight allocation, influenced by factors such as equipment fixity, monitoring necessity, and technical constraints. The degree of optimization of surrounding nodes reflects the optimization effect of the allocation operation on adjacent monitored nodes; allocation schemes with more significant optimization effects have higher value. In the GPU computing cluster area of ​​the data center, the resource importance level is very high, but the allocateability is relatively low, while the surrounding general-purpose server area has a higher allocateability and a significant degree of optimization. The contribution of each factor is quantified using a value assessment model: resource importance level has a weight of 0.4, allocateability has a weight of 0.35, and the degree of optimization of surrounding nodes has a weight of 0.25. Based on the comprehensive value assessment results, the allocation intensity level is determined, divided into three levels: strong allocation, medium allocation, and slight allocation. Emphasis allocation is suitable for areas with high redundancy and high allocation value; medium allocation is suitable for areas with medium redundancy and controllable allocation risk; and slight allocation is suitable for areas with slight redundancy and requiring fine-tuning.

[0061] The remaining weights of surrounding monitoring nodes are redistributed based on the allocation intensity, forming a remaining weight allocation. The redistribution process is guided by allocation intensity levels, emphasizing that redundant weights in redundant areas can be significantly transferred to underweight surrounding nodes. A weight transfer algorithm distributes excess weights from redundant areas to low-weight surrounding nodes according to a certain proportion, with the transfer ratio determined based on the spatial distance and functional similarity between nodes. In the network equipment area of ​​the data center, some redundant weights of core switch nodes are transferred to edge switch nodes, resulting in a more balanced weight distribution across the entire network equipment area. The impact of weight transfer on monitoring performance is considered to ensure that weight allocation does not degrade monitoring quality in critical areas. An iterative optimization method is used to gradually adjust the weight allocation scheme, evaluating the allocation effect and making fine-tuning in each iteration. The redistributed weight results are compiled into a remaining weight allocation table, which includes fields such as node identifier, original weight, allocated weight, final weight, and allocation source.

[0062] A resource allocation execution sequence is established using surplus weight allocation. Based on the allocation results in the surplus weight allocation table, weight transfer tasks are transformed into specific execution operation sequences. According to the complexity and scope of impact of weight allocation, allocation tasks are prioritized according to execution difficulty, with simpler and lower-risk tasks being executed first. Single-node weight adjustment operations have low complexity, multi-node collaborative allocation operations have high complexity, and cross-regional weight transfer operations have the highest complexity. Considering resource constraints and time window limitations during the allocation process, a reasonable execution time schedule is formulated to avoid the allocation operations affecting normal monitoring operations. In the storage system area of ​​the data center, since storage devices typically perform data backup and synchronization at night, weight allocation in this area is scheduled for execution during low-load periods at night. The output results of the weight transfer algorithm are sorted according to time priority: core node weight adjustment has the highest priority, followed by auxiliary node weight adjustment, and boundary node weight adjustment has the lowest priority. The interdependencies between allocation tasks are analyzed to determine the execution order and parallelization possibilities.

[0063] A surplus allocation table is generated through resource allocation execution sequences. The defined resource allocation execution sequences are organized into a structured surplus allocation table, which records detailed execution information for each allocation task. The surplus allocation table includes key fields such as allocation task number, source node identifier, target node identifier, allocation weight, execution time, execution status, and completion marker. Allocation tasks are sorted by execution time to form a chronological allocation execution plan. In the cooling system area of ​​the data center, because the allocation of precision air conditioning equipment involves changes in environmental parameters, allocation tasks in this area are scheduled to be executed during system maintenance windows, and the allocation table indicates the specific maintenance time windows. The completeness and consistency of the surplus allocation table are analyzed to ensure that all weight redundancies are reasonably allocated and all weight deficient nodes are appropriately supplemented. The surplus allocation table is integrated with the resource management module of the monitoring system to support automated execution and status tracking of allocation tasks.

[0064] A margin allocation table is used to optimize monitoring paths and form optimized transmission paths. Based on the allocation results of the margin allocation table, the connection relationships and data transmission paths between monitoring nodes are re-analyzed. The optimal data transmission path after weight allocation is identified through network topology optimization methods, with the optimization objective being to minimize data transmission costs while ensuring monitoring effectiveness. A combination of shortest path and maximum flow algorithms is used to find the optimal transmission path from the data source node to the data processing center. The impact of weight allocation on transmission path capacity is considered; higher-weight nodes generate more monitoring data and require greater transmission bandwidth. In the server cluster area of ​​the data center, the weight of some edge server nodes increases after weight allocation, and the corresponding data transmission paths need to be expanded to support the increased data traffic. The load balancing characteristics of the transmission paths are analyzed to avoid situations where some paths are overloaded while others are idle. Backup transmission paths are added through path redundancy analysis to improve the reliability and fault tolerance of the monitoring network. The optimized transmission paths are classified according to priority, with the primary transmission path undertaking the main data transmission task, and backup transmission paths activated when the primary path fails.

[0065] To optimize transmission paths, sequence calibration was performed to construct differentiated monitoring sequences. The monitoring sequences were recalibrated and reordered using the topology and transmission characteristics of the optimized transmission paths. Path analysis determined the importance level of each monitoring node in the transmission network, with nodes on critical transmission paths having higher sequence priority. Differentiated monitoring sequences were categorized into three levels based on node importance: critical node sequences (10-second monitoring interval, full parameter monitoring), important node sequences (30-second monitoring interval, key parameter monitoring), and ordinary node sequences (60-second monitoring interval, threshold monitoring). The activation sequence and data collection order of monitoring nodes were determined based on the data flow direction and transmission delay characteristics of the transmission paths. In the core network area of ​​the data center, the core switch node, being the convergence point of multiple transmission paths, was designated as the primary monitoring node in the sequence, with the highest priority. Considering the collaborative and data dependencies between monitoring nodes, nodes with strong collaborative relationships were organized into continuous segments of the monitoring sequence. A specific monitoring time window was assigned to each node in the monitoring sequence using a time calibration method, taking into account factors such as node weight, data processing capacity, and transmission delay. The calibrated monitoring nodes are organized into differentiated monitoring sequences according to priority and time order. The sequences are reflected in three aspects: differences in monitoring frequency, differences in parameter range, and differences in response speed.

[0066] In some embodiments, generating an energy consumption and carbon emission analysis matrix by performing matrix operations based on the differentiated monitoring sequence includes: identifying redundant data elements based on the differentiated monitoring sequence; performing redundancy utilization analysis on the redundant data elements to obtain a redundancy value coefficient; obtaining a simplified processing recursive chain according to the redundancy value coefficient; and generating the energy consumption and carbon emission analysis matrix by performing matrix operations through the simplified processing recursive chain.

[0067] Data redundancy elements are identified based on differentiated monitoring sequences. The data characteristics generated by each monitoring node in the differentiated monitoring sequences are analyzed to identify data elements with information overlap and content redundancy. Data content from different monitoring nodes is compared using data similarity analysis; when the correlation coefficient between two nodes exceeds 0.8, data redundancy is considered to exist. Information entropy analysis is used to quantify the information content of the data; data elements with lower information entropy typically contain more redundant information. In the storage array area of ​​the data center, due to the similar operating modes and consistent environmental conditions of multiple disk racks, the monitoring data in this area exhibits high similarity, with generally low information entropy values. Time series analysis is used to identify repetitive patterns and periodic components in the data; repetitive patterns correspond to temporal redundancy, and periodic components correspond to structural redundancy. The spatial distribution characteristics of data redundancy elements are analyzed; adjacent monitoring nodes are more likely to generate spatially redundant data due to similar environmental conditions. Cluster analysis is used to group data elements with similar redundancy characteristics; data elements within each redundancy group are substitutable. In the network equipment area of ​​the data center, due to the similar functions of the switching equipment, the network traffic and power consumption data generated by multiple monitoring nodes in this area exhibit significant redundancy characteristics. The identified redundant data elements are classified according to their degree of redundancy. High-redundancy elements can be greatly simplified, medium-redundancy elements need to be selectively retained, and low-redundancy elements are basically left as is.

[0068] Redundancy utilization analysis is performed on redundant data elements to obtain redundancy value coefficients. The value of identified redundant data elements is assessed to analyze their potential utilization value in monitoring and analysis. The reliability and accuracy of redundant data are determined through data quality assessment; high-quality redundant data has higher utilization value. The applicability of redundant data in different application scenarios is analyzed; some redundant data, while having low value in the current analysis, may play an important role in other analytical tasks. In GPU computing clusters in data centers, due to the highly similar load patterns of multiple GPU servers, this redundant load data can be used for load prediction and resource scheduling optimization. The representativeness of redundant data is determined through statistical characteristic analysis; highly representative redundant data can serve as typical samples of similar data. An information value assessment method is used to quantify the information contribution of redundant data, reflecting the degree of influence of redundant data on the overall analysis results. The formula for calculating the redundancy value coefficient is V = 0.3 × Q + 0.25 × R + 0.25 × A + 0.2 × I, where V is the redundancy value coefficient, Q is the data quality score, R is the representativeness, A is the applicability score, and I is the information contribution. The redundancy value coefficient ranges from 0 to 1, with a higher coefficient indicating greater utilization value of redundant data. In the data center cooling system area, although there is redundancy in the operating data of multiple precision air conditioners, the redundancy value coefficient is set to a high level because this data is important for temperature control analysis.

[0069] A simplified processing recursive chain is obtained based on the redundancy value coefficient. According to the magnitude of the redundancy value coefficient, corresponding data simplification strategies and recursive processing flows are formulated. For data elements with high redundancy value coefficients, a data fusion method is used to merge multiple redundant data into representative data. For data elements with low redundancy value coefficients, a data sampling method is used to select and retain a portion of representative samples. The data simplification process is handled through a recursive algorithm, with the recursive chain starting from the data with the highest redundancy and gradually progressing to data with lower redundancy. In the recursive processing, each simplification operation affects the next processing decision, forming a layered recursive processing chain. In the server room area of ​​the data center, for similar monitoring data from multiple web servers, the recursive chain first processes the most similar data pairs, and then gradually expands to data combinations with lower similarity. Through iterative processing of the recursive chain, the degree of data redundancy is gradually reduced while retaining core information features.

[0070] An energy consumption and carbon emission analysis matrix is ​​generated by performing matrix operations using a simplified recursive chain. The output of the simplified recursive chain is used as the input data for matrix operations to ensure computational efficiency and accuracy. The simplified monitoring data is organized into a standard matrix format according to spatial location and parameter type, with rows corresponding to monitoring nodes and columns corresponding to monitoring parameters. Data fusion and analysis are achieved through basic matrix operations, including matrix addition for data aggregation, matrix multiplication for weight calculation, and matrix transpose for dimensionality transformation. A block matrix operation method is used to process large-scale monitoring data, decomposing large matrices into several smaller matrices for parallel computation to improve computational efficiency. In the hybrid computing area of ​​the data center, which contains different types of equipment such as CPU servers, GPU servers, and storage devices, the block matrix method is used to process monitoring data from different equipment types separately. The main features and correlation patterns in the data are extracted through eigenvalue decomposition and singular value decomposition of the matrix to identify the dominant relationship between energy consumption and carbon emissions. The processing results are organized into an energy consumption and carbon emission analysis matrix, which includes sub-matrices for energy intensity, carbon emission density, efficiency assessment, and correlation analysis.

[0071] Statistical evaluation of the energy consumption and carbon emission analysis matrix generates monitoring configuration parameters. Mean analysis of the energy intensity sub-matrix determines the average energy consumption level of each region in the data center, identifying high- and low-energy-consumption areas. Variance analysis of the carbon emission density sub-matrix assesses the stability and variability of carbon emissions in each region; regions with higher variance indicate more drastic carbon emission fluctuations, requiring more frequent monitoring. In the data center's web server cluster, analysis of the carbon emission density sub-matrix reveals that the carbon emission variance in this region is significantly higher than that in the storage region, necessitating a high-frequency monitoring configuration. Statistical analysis of the efficiency assessment sub-matrix identifies regions with low energy utilization efficiency, requiring emission reduction optimization monitoring modes. Correlation analysis of the correlation sub-matrix identifies the degree of correlation and propagation relationships of carbon emissions between different regions. Based on the statistical evaluation results of the four sub-matrixes, monitoring configuration parameters are generated, including key parameters such as monitoring frequency configuration, data transmission configuration, storage strategy configuration, and alarm threshold configuration. The monitoring frequency configuration comprehensively considers the change frequencies of the energy consumption intensity and carbon emission density sub-matrixes to determine the sampling interval. Data transmission configuration prioritizes data based on the evaluation results of the efficiency assessment submatrix. Data from areas with high efficiency assessment values ​​is transmitted in real time, while data from areas with low efficiency assessment values ​​is transmitted in batches. The generated monitoring configuration parameters are then applied to the actual monitoring system to ultimately achieve intelligent monitoring of data center energy consumption and carbon emissions.

[0072] To implement the above-described method embodiments, a smart monitoring method for data center energy consumption and carbon emissions is proposed to achieve the corresponding functions and technical effects. See also... Figure 2 , Figure 2This diagram illustrates a structural block diagram of a data center energy consumption and carbon emission intelligent monitoring system 200 according to an embodiment of this application. For ease of explanation, only the parts relevant to this embodiment are shown. The data center energy consumption and carbon emission intelligent monitoring system 200 provided in this embodiment includes: Data acquisition module 201 is used to monitor energy consumption monitoring data and environmental parameter data in the data center operating environment, perform time-series analysis on the energy consumption monitoring data to identify energy consumption patterns, and generate graded quality parameters based on the quality requirements of the energy consumption monitoring data. Marginal analysis module 202 is used to perform marginal utility evaluation based on the energy consumption pattern and the graded quality parameters to determine the precision configuration boundary, perform impact intensity evaluation on the precision configuration boundary to generate a carbon source intensity index, and use spatial mapping processing to convert the carbon source intensity index into an impact coverage area. The propagation analysis module 203 is used to analyze the change trajectory of the environmental parameter data to generate a carbon propagation network, perform frequency analysis on the carbon propagation network to obtain key propagation paths, map the key propagation paths to carbon mobility indicators, and reconstruct the monitoring network based on the carbon mobility indicators to form a path optimization table. Range control module 204 is used to assess the abnormal propagation range of the affected coverage area to determine the monitoring extension area, perform dynamic range adjustment on the monitoring extension area to generate range adjustment parameters, integrate the range adjustment parameters with the path optimization table to construct an adaptive monitoring configuration, and identify emission reduction potential areas based on the adaptive monitoring configuration; The matrix processing module 205 is used to decompose the emission reduction potential area into spatial monitoring nodes, allocate utility weights to the spatial monitoring nodes to generate differentiated monitoring sequences, perform matrix operations based on the differentiated monitoring sequences to generate an energy consumption and carbon emission analysis matrix, perform statistical evaluation on the energy consumption and carbon emission analysis matrix to generate monitoring configuration parameters, and complete intelligent monitoring of data center energy consumption and carbon emission.

[0073] The aforementioned intelligent data center energy consumption and carbon emission monitoring system 200 can implement the intelligent data center energy consumption and carbon emission monitoring method described in the above-described method embodiments. The options described in the above method embodiments are also applicable to this embodiment and will not be detailed here. The remaining content of this application's embodiments can be referred to the content of the above method embodiments, and will not be repeated in this embodiment.

[0074] The purpose of the above embodiments is to reproduce and derive the technical solution of the present invention by way of example, and to fully describe the technical solution, purpose and effect of the present invention. The purpose is to enable the public to have a more thorough and comprehensive understanding of the disclosure of the present invention, and not to limit the scope of protection of the present invention.

[0075] The above embodiments are not an exhaustive list based on the present invention, and there may be many other embodiments not listed. Any substitutions and improvements made without departing from the concept of the present invention are within the protection scope of the present invention.

Claims

1. A method for intelligent monitoring of energy consumption and carbon emissions in data centers, characterized in that, include: Monitor energy consumption data and environmental parameter data within the data center operating environment, perform time-series analysis on the energy consumption data to identify energy consumption patterns, and generate graded quality parameters based on the quality requirements of the energy consumption data; Based on the energy consumption pattern and the graded quality parameters, a marginal utility assessment is performed to determine the precision configuration boundary. An impact intensity assessment is then performed on the precision configuration boundary to generate a carbon source intensity index. Spatial mapping processing is then used to convert the carbon source intensity index into an impact coverage area. The change trajectory of the environmental parameter data is analyzed to generate a carbon propagation network. Frequency analysis is performed on the carbon propagation network to obtain key propagation paths. The key propagation paths are mapped to carbon mobility indicators. Based on the carbon mobility indicators, the monitoring network is reconstructed to form a path optimization table. Assess the abnormal propagation range of the affected coverage area to determine the monitoring extension area, perform dynamic range adjustment on the monitoring extension area to generate range adjustment parameters, fuse the range adjustment parameters with the path optimization table to construct an adaptive monitoring configuration, and identify emission reduction potential areas based on the adaptive monitoring configuration; The emission reduction potential area is decomposed into spatial monitoring nodes. The spatial monitoring nodes are assigned utility weights to generate differentiated monitoring sequences. Matrix operations are performed on the differentiated monitoring sequences to generate an energy consumption and carbon emission analysis matrix. The energy consumption and carbon emission analysis matrix is ​​statistically evaluated to generate monitoring configuration parameters, thus completing the intelligent monitoring of energy consumption and carbon emission in the data center.

2. The method according to claim 1, characterized in that, The step of determining the precision configuration boundary by performing marginal utility evaluation based on the energy consumption pattern and the graded quality parameters includes: Identify the peak frequency band of consumption intensity for the aforementioned energy consumption pattern; Utility correlation points are generated by correlating the peak frequency band of consumption intensity with the graded quality parameters. The aforementioned utility correlation points are used to stimulate changes in the distribution of precision requirements, thereby generating a precision requirement domain. The accuracy configuration boundary is determined by the accuracy requirement domain.

3. The method according to claim 1, characterized in that, The process of converting the carbon source intensity index into an impact coverage area using spatial mapping includes: Based on the carbon source intensity index, spatial mapping is performed to identify differences in mapping density, generating dense and sparse mapping regions. The sparse mapping region is used to perform influence range inference processing to form an extended inference region; The extended inference region and the dense mapping region are spatially fused to generate a complete coverage segment; Spatial optimization is performed on the complete coverage segment to construct the affected coverage area.

4. The method according to claim 1, characterized in that, The step of performing frequency analysis on the carbon propagation network to obtain key propagation paths includes: Obtain the propagation frequency characteristics of the carbon propagation network; Based on the propagation frequency characteristics, low-frequency propagation regions are identified and low-frequency region parameters are generated; The propagation signal is passively adjusted using the low-frequency region parameters to generate an regulated propagation signal. The key propagation path is identified by adjusting the propagation signal.

5. The method according to claim 1, characterized in that, The assessment of the abnormal propagation range of the affected coverage area to determine the monitoring extension area includes: Based on the aforementioned area of ​​impact coverage, measure the propagation range anomalies to identify low-anomaly regions; The low-anomaly region is used as a stability reference to form a reference quality distribution; The range value is derived using the reference mass distribution to generate the derived range matrix; Spatial mapping is performed on the derived range matrix to determine the monitoring extension area.

6. The method according to claim 1, characterized in that, The step of assigning utility weights to the spatial monitoring nodes to generate differentiated monitoring sequences includes: For the aforementioned spatial monitoring nodes, utility weight allocation is performed to identify weight redundancy regions; Based on the aforementioned weighted redundancy region, an adjustable margin is determined to form a margin allocation table; The aforementioned margin allocation table is used to optimize the monitoring path and form an optimized transmission path; Sequence calibration is performed on the optimized transmission path to construct a differentiated monitoring sequence.

7. The method according to claim 1, characterized in that, The step of generating an energy consumption and carbon emission analysis matrix by performing matrix operations based on the differentiated monitoring sequences includes: Data redundancy elements are identified based on the differentiated monitoring sequences; Redundancy utilization analysis is performed on the aforementioned data redundancy elements to obtain redundancy value coefficients; A simplified processing recursive chain is obtained according to the aforementioned redundancy value coefficient; The simplified processing recursive chain is used to perform matrix operations to generate an energy consumption and carbon emission analysis matrix.

8. The method according to claim 2, characterized in that, The step of using the utility correlation points to trigger changes in the precision requirement distribution to generate a precision requirement domain includes: Based on the utility correlation point analysis accuracy configuration status, a configuration margin table is generated; A density optimization curve is generated based on the distribution changes in the accuracy requirements of the configuration margin table. The density optimization curve is used to identify advantageous configuration paths. Spatial mapping is performed on the advantageous configuration path to generate the accuracy requirement domain.

9. The method according to claim 6, characterized in that, The step of determining the adjustable margin based on the weighted redundancy region to form a margin allocation table includes: The allocation intensity is determined based on the weighted redundant area assessment of the surplus allocation value, wherein the surplus allocation value includes the resource importance level, allocation degree, and optimization degree of the surrounding nodes of the redundant area; The remaining weights of the surrounding monitoring nodes are redistributed according to the allocation intensity to form a remaining weight allocation. The resource allocation execution sequence is determined using the aforementioned residual weight allocation; The resource allocation execution sequence forms a surplus allocation table.

10. A smart monitoring system for energy consumption and carbon emissions in a data center, characterized in that, include: The data acquisition module is used to monitor energy consumption monitoring data and environmental parameter data in the data center operating environment, perform time-series analysis on the energy consumption monitoring data to identify energy consumption patterns, and generate graded quality parameters based on the quality requirements of the energy consumption monitoring data. The marginal analysis module is used to perform marginal utility evaluation based on the energy consumption pattern and the graded quality parameters to determine the precision configuration boundary, perform impact intensity evaluation on the precision configuration boundary to generate a carbon source intensity index, and use spatial mapping processing to convert the carbon source intensity index into an impact coverage area. The propagation analysis module is used to analyze the change trajectory of the environmental parameter data to generate a carbon propagation network, perform frequency analysis on the carbon propagation network to obtain key propagation paths, map the key propagation paths to carbon mobility indicators, and reconstruct the monitoring network based on the carbon mobility indicators to form a path optimization table. The range control module is used to assess the abnormal propagation range of the affected coverage area, determine the monitoring extension area, perform dynamic range adjustment on the monitoring extension area to generate range adjustment parameters, integrate the range adjustment parameters with the path optimization table to construct an adaptive monitoring configuration, and identify emission reduction potential areas based on the adaptive monitoring configuration. The matrix processing module is used to decompose the emission reduction potential area into spatial monitoring nodes, allocate utility weights to the spatial monitoring nodes to generate differentiated monitoring sequences, perform matrix operations based on the differentiated monitoring sequences to generate an energy consumption and carbon emission analysis matrix, perform statistical evaluation on the energy consumption and carbon emission analysis matrix to generate monitoring configuration parameters, and complete intelligent monitoring of energy consumption and carbon emission in the data center.

Images