Multi-dimensional coupling machine room energy-saving optimization control method and system

Abstract

The invention relates to the field of machine room energy-saving control, and discloses a multi-dimensional coupled machine room energy-saving optimization control method and system, and the method comprises the steps: obtaining and preprocessing the multi-dimensional operation data of a machine room, constructing an energy consumption evaluation model according to the preprocessed data, evaluating the current energy consumption situation of the machine room, and recognizing the current energy consumption state; energy flow distribution is analyzed according to the energy consumption state, and a high-loss link and a low-efficiency area are accurately determined; aiming at the determined problem area, constructing an inter-equipment multi-dimensional coupling model; and finally, inputting real-time data into the model, outputting prediction data, generating an energy-saving optimization control strategy according to the prediction data, executing the energy-saving optimization control strategy, and adjusting the operation state of the equipment, thereby accurately positioning an energy consumption problem area according to the real-time operation condition of the machine room, constructing an equipment coupling relation model, and dynamically generating and executing the energy-saving strategy. Finally, the purposes of reducing the overall energy consumption of the machine room and improving the energy utilization efficiency and the equipment operation stability are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of machine room energy-saving control, and more particularly to a multi-dimensional coupling machine room energy-saving optimization control method and system. BACKGROUND

[0002] With the continuous expansion of data centers and the exponential growth of computing power demand, the energy consumption problem of machine rooms has become a core bottleneck restricting the sustainable development of the industry. Traditional methods mainly rely on simple control logic based on static thresholds, such as starting and stopping air conditioning units according to fixed temperature points. With the development of technology, existing technologies have introduced modeling and evaluation based on historical data, identified high-energy-consuming devices by constructing energy consumption models, and made strategy adjustments according to expert experience or preset rule libraries, realizing preliminary optimization from passive response to data analysis-based optimization and improving the fine level of machine room energy management.

[0003] However, the existing technology still has the following shortcomings: First, the existing technology constructs an energy consumption model based on historical data, which can only identify high-energy-consuming devices and cannot quantify the coupling strength between devices, resulting in locally effective but globally deteriorating optimization strategies; Second, the modeling based on historical data in the existing technology lacks real-time prediction capability and cannot respond to sudden changes in working conditions (such as sudden computing power peaks or sudden temperature rises), resulting in a lag in control strategies and a single optimization goal that does not consider multi-objective coordination, which may cause devices to be frequently started and stopped or overloaded due to excessive optimization; Third, the existing technology lacks a feedback mechanism, with fixed model parameters that cannot adapt to device aging or topology changes (such as the addition of servers or the replacement of air conditioner models), resulting in a significant decrease in control accuracy over a long period of operation, and the expert rule library relies on human experience, which is difficult to cover all working conditions and is prone to misoperation. SUMMARY

[0004] In order to overcome the above-mentioned defects of the prior art, the present application provides a multi-dimensional coupling machine room energy-saving optimization control method and system to solve the problems existing in the background art.

[0005] The present application provides the following technical solutions: a multi-dimensional coupling machine room energy-saving optimization control method, comprising: S1: acquiring multi-dimensional running data of a machine room and preprocessing the collected multi-dimensional running data; S2: constructing a machine room energy consumption evaluation model based on the preprocessed multi-dimensional running data, evaluating the current energy consumption status of the machine room based on the machine room energy consumption evaluation model, and identifying the current energy consumption status of the machine room; S3: analyzing the energy flow distribution in the machine room according to the identified current energy consumption status, determining the high-loss links and low-efficiency areas in the energy flow; S4: constructing a multi-dimensional coupling model between devices in the machine room according to the determined high-loss link and low-efficiency area; S5: inputting real-time multi-dimensional running data into the multi-dimensional coupling model to output predicted multi-dimensional running data, generating an energy-saving optimization control strategy according to the predicted multi-dimensional running data, and executing the energy-saving optimization control strategy to adjust the running state of the device.

[0006] Preferably, S1 collects multi-dimensional running data in the machine room in real time by deploying a sensor network, an IT device intelligent interface, and a power monitoring system in the machine room, and the multi-dimensional running data includes environmental temperature, environmental humidity, device power consumption, server utilization, device temperature, device vibration frequency, network traffic, and power supply parameters. The specific operations of preprocessing the collected multi-dimensional running data include data alignment, data cleaning, data conversion, and data normalization processing.

[0007] Preferably, S2 extracts feature parameters affecting the energy consumption status of the machine room from the multi-dimensional running data by performing correlation analysis and feature extraction on the preprocessed multi-dimensional running data. Based on the extracted feature parameters, a machine room energy consumption evaluation model is constructed. A theoretical optimal energy consumption value is output by the model by inputting a real-time feature parameter sequence into the constructed machine room energy consumption evaluation model. By comparing the actual total energy consumption value of the machine room with the theoretical optimal energy consumption value, an energy efficiency deviation index is calculated as a quantitative indicator for evaluating the current energy consumption status of the machine room. The calculated energy efficiency deviation index is compared with a plurality of preset energy consumption state threshold intervals to identify and output the current energy consumption state of the machine room, including high-efficiency running state, normal running state, low-efficiency running state, and abnormal high-energy consumption state. If it is identified that the machine room is in a low-efficiency running state or an abnormal high-energy consumption state, an analysis process of the energy flow distribution in the machine room is triggered; if it is identified that the machine room is in a high-efficiency running state or a normal running state, the monitoring and evaluation of the current energy consumption status of the machine room is continued.

[0008] Preferably, S3 further analyzes the energy flow distribution in the machine room when it is identified that the machine room is in a low-efficiency running state or an abnormal high-energy consumption state, and the specific analysis process includes: Based on the collected multi-dimensional running data, real-time current, voltage, and power factor data of each power branch circuit, and energy input and output data of each functional area are further collected, and the data is classified and integrated according to device type, functional area, and time dimension to form a structured energy data set. Based on the structured energy dataset, an energy flow network model is constructed, which takes the mains input as the source node, each electrical equipment as the intermediate node, and the final energy dissipation form as the sink node. The energy flow network model is a weighted directed graph, wherein the weight of the edge represents the size of the energy flow, and the flow of energy in the energy flow network model and the loss distribution of each node and edge are calculated. Based on the energy flow network model, a multi-level loss analysis is performed, which includes: Device-level analysis: used for analyzing high-loss links, by calculating the real-time energy utilization rate of each device and comparing it with the preset device efficiency benchmark value, the energy utilization deviation value is obtained, when the energy utilization deviation value exceeds the preset deviation threshold, the compared device is identified and labeled as an inefficient device, the preset device efficiency benchmark value is determined based on the device design parameters and historical optimal operation data, and is dynamically modified every quarter according to the device maintenance record and working condition change; Line-level analysis: the real-time resistance loss of each power branch line is calculated using Joule's law, and the real-time resistance loss is compared with the line benchmark resistance loss value calculated according to the line design parameters and rated working condition, when the real-time resistance loss is higher than the line benchmark resistance loss value, the compared line is located, and the located line is labeled as a high-loss line; Regional-level analysis: the regional energy efficiency index of each functional area is calculated, and the regional energy efficiency index is compared with the preset regional energy efficiency benchmark value, if the regional energy efficiency index is lower than the preset regional energy efficiency benchmark value, the compared functional area is located, and the located area is labeled as a low-energy efficiency area; The located inefficient devices and high-loss lines are used to determine the high-loss links in the energy flow, and the located low-energy efficiency areas are used to determine the inefficient areas in the energy flow.

[0009] Preferably, the specific process of S4 for constructing a multi-dimensional coupling model according to the determined high-loss links and inefficient areas includes: Based on the physical topology, spatial position and operation logic between devices, the coupling relationship including energy flow and thermal dimension is defined, a multi-dimensional coupling network model is constructed with devices as nodes and the coupling relationship as edges, and the coupling strength of each edge is quantified using historical operation data, the determined inefficient devices, high-loss lines and inefficient areas are loaded into the multi-dimensional coupling network model as abnormal nodes.

[0010] Preferably, S5 generates and executes an energy-saving optimization control strategy according to the multi-dimensional coupling model, which includes: The real-time multi-dimensional operation data is input into the multi-dimensional coupling network model for simulation deduction, so that predicted multi-dimensional operation data of equipment operation state and energy consumption response in a future time window is output; Based on the output predicted multi-dimensional operation data and a preset energy-saving target, an optimal control parameter combination is solved through an optimization algorithm, the optimal control parameter combination is converted into an energy-saving optimization control strategy for specific equipment, and the energy-saving optimization control strategy is executed to dynamically adjust the operation state of the related equipment. The adjusted actual multi-dimensional operation data is continuously monitored, the deviation between the actual multi-dimensional operation data and the predicted multi-dimensional operation data is compared, the multi-dimensional coupling network model is parameter-modified, and the actual multi-dimensional operation data is used as input data and applied to the parameter-modified multi-dimensional coupling network model for the next round of optimization.

[0011] To achieve the above object, the present application provides the following technical scheme: a multi-dimensional coupling computer room energy-saving optimization control system implementing the above multi-dimensional coupling computer room energy-saving optimization control method, comprising: A multi-dimensional data acquisition module acquires multi-dimensional operation data of the computer room and pre-processes the acquired multi-dimensional operation data. A computer room energy consumption state recognition module constructs a computer room energy consumption evaluation model according to the pre-processed multi-dimensional operation data, evaluates the current energy consumption state of the computer room based on the computer room energy consumption evaluation model, and recognizes the current energy consumption state of the computer room. An energy flow distribution analysis module analyzes the energy flow distribution in the computer room according to the recognized current energy consumption state, and determines the high-loss link and low-efficiency area in the energy flow. A multi-dimensional coupling model construction module constructs a multi-dimensional coupling model between devices in the computer room according to the determined high-loss link and low-efficiency area. A control strategy generation and adjustment module inputs real-time multi-dimensional operation data into the multi-dimensional coupling model, outputs predicted multi-dimensional operation data, generates an energy-saving optimization control strategy according to the predicted multi-dimensional operation data, and executes the energy-saving optimization control strategy to adjust the operation state of the equipment.

[0012] The technical effects and advantages of the present application are as follows: By constructing a multi-dimensional coupling network model, the strength of energy transmission and thermal influence between devices can be determined by taking devices as nodes and coupling relationships as edges, and the high-loss link and low-efficiency area can be located by combining the energy flow network model, so that precise diagnosis from the "device level" to the "system level" is realized, and global energy efficiency deterioration caused by local optimization is avoided.

[0013] By inputting real-time data into the multi-dimensional coupling model, the device state and energy consumption response in the future time window are simulated, sudden changes in working conditions are responded to in advance, based on the prediction results, the optimal control parameter combination is solved through an optimization algorithm, the device operation parameters are constrained while meeting the energy efficiency target, and over-optimization is avoided to cause frequent start-stop or overload of the device.

[0014] By continuously monitoring the deviation between actual running data and predicted values, the model parameters are automatically corrected to adapt to device aging or topology changes, the identified inefficient devices, high-loss lines and inefficient areas are loaded as abnormal nodes into the multi-dimensional coupling model, and the coupling relationship is dynamically adjusted to avoid abnormal working conditions affecting global optimization and reduce the risk of manual intervention and misoperation. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 The method steps of the present application.

[0016] Figure 2 The system structure diagram of the present application. DETAILED DESCRIPTION

[0017] The technical solutions in the present application will be described clearly and completely below in combination with the drawings in the present application, and additionally, the forms of each structure described in the following embodiments are only examples, and the multi-dimensional coupling data center energy-saving optimization control method and system involved in the present application are not limited to each structure described in the following embodiments. All other embodiments obtained by those skilled in the art without making creative efforts belong to the scope of protection of the present application.

[0018] As Figure 1 shown, the present embodiment provides a multi-dimensional coupling data center energy-saving optimization control method, comprising: S1: acquiring multi-dimensional running data of the data center, and preprocessing the collected multi-dimensional running data.

[0019] In the present embodiment, S1 acquires multi-dimensional running data in the data center in real time by deploying a sensor network, an IT device intelligent interface and a power monitoring system in the data center, and the multi-dimensional running data includes environmental temperature, environmental humidity, device power consumption, server utilization, device temperature, device vibration frequency, network traffic and power supply parameters. The specific operation of preprocessing the collected multi-dimensional running data includes data alignment, data cleaning, data conversion and data normalization processing.

[0020] S2: constructing a data center energy consumption evaluation model according to the preprocessed multi-dimensional running data, evaluating the current energy consumption status of the data center based on the data center energy consumption evaluation model, and identifying the current energy consumption state of the data center.

[0021] In this embodiment, S2 extracts feature parameters affecting the energy consumption status of the machine room from the multi-dimensional running data by performing correlation analysis and feature extraction on the pre-processed multi-dimensional running data. Based on the extracted feature parameters, a machine room energy consumption evaluation model is constructed. The real-time feature parameter sequence is input into the constructed machine room energy consumption evaluation model, and a theoretical optimal energy consumption value is output by the model. By comparing the actual total energy consumption value of the machine room with the theoretical optimal energy consumption value, an energy efficiency deviation index is calculated as a quantitative indicator for evaluating the current energy consumption status of the machine room. The calculated energy efficiency deviation index is compared with a plurality of preset energy consumption state threshold intervals to identify and output the current energy consumption state of the machine room, including: high-efficiency running state, normal running state, low-efficiency running state and abnormal high-energy consumption state. If it is identified that the machine room is in a low-efficiency running state or an abnormal high-energy consumption state, an analysis process of the energy flow distribution in the machine room is triggered; if it is identified that the machine room is in a high-efficiency running state or a normal running state, the monitoring and evaluation of the current energy consumption status of the machine room is continued.

[0022] It needs to be specifically explained that the correlation analysis and feature extraction on the pre-processed multi-dimensional running data can be performed by using Pearson correlation coefficient, mutual information or principal component analysis (PCA) method, etc. to quantify the correlation between each dimension data and the total energy consumption, and to filter out the feature parameters that have significant impact on energy consumption, such as total power of IT equipment, average running temperature of equipment, energy efficiency ratio of refrigeration system, etc. Based on the extracted feature parameters, a machine room energy consumption evaluation model is constructed, which is preferably a machine learning model trained based on historical optimal running data, such as random forest, gradient boosting decision tree or neural network, and the target is to learn the lowest energy consumption level that can be theoretically achieved under specific IT load and environmental conditions, i.e. to establish a nonlinear mapping relationship from feature parameters to theoretical optimal energy consumption value; the real-time collected and processed feature parameter sequence is input into the trained machine room energy consumption evaluation model, and the model will output a theoretical optimal energy consumption value under the current working condition. By comparing the actual total energy consumption value of the machine room with the theoretical optimal energy consumption value, an energy efficiency deviation index is calculated, and the calculation formula is wherein represents the actual total energy consumption value of the machine room, represents the theoretical optimal energy consumption value, and the energy efficiency deviation index directly reflects the deviation degree of the current energy consumption of the machine room relative to the ideal state. The larger the energy efficiency deviation index value is, the lower the energy efficiency is, and the more serious the energy waste is. The calculated energy efficiency deviation index is compared with a plurality of preset energy consumption state threshold intervals to identify and output the current energy consumption state of the machine room, for example, the threshold interval is set as: when , it is determined that the current energy consumption state of the machine room is “high-efficiency running state”; when At that time, the current energy consumption status of the computer room is determined to be "normal operating status"; When the current energy consumption status of the computer room is determined to be "inefficient operation status", when At that time, the current energy consumption status of the computer room is determined to be "abnormally high energy consumption status".

[0023] S3: Based on the identified current energy consumption status, analyze the energy flow distribution in the computer room and determine the high-loss links and low-efficiency areas in the energy flow.

[0024] In this embodiment, when S3 detects that the computer room is in an inefficient operating state or an abnormally high energy consumption state, it immediately performs further analysis on the energy flow distribution within the computer room. The specific analysis process includes: Based on the collected multi-dimensional operational data, further collect real-time current, voltage, and power factor data of each power branch circuit, as well as energy input and output data of each functional area. Then, classify and integrate the data according to equipment type, functional area, and time dimension to form a structured energy dataset. Based on a structured energy dataset, an energy flow network model is constructed with mains power input as the source node, each electrical device as the intermediate node, and the final energy dissipation form as the sink node. The energy flow network model is a weighted directed graph, where the weight of the edge represents the magnitude of energy flow. The flow of energy and the loss distribution of energy in each node and edge of the energy flow network model are calculated. Based on the energy flow network model, multi-level loss analysis is performed, which includes: Equipment-level analysis: Used to analyze high-loss processes. It calculates the real-time energy utilization rate of each piece of equipment and compares it with the preset equipment efficiency benchmark value to obtain the energy utilization deviation value. When the energy utilization deviation value exceeds the preset deviation threshold, the compared equipment is identified and marked as inefficient equipment. The preset equipment efficiency benchmark value is determined based on the equipment design parameters and historical best operating data, and is dynamically corrected every quarter based on equipment maintenance records and changes in operating conditions. Line-level analysis: Joule's law is used to calculate the real-time resistance loss of each power branch line, and the real-time resistance loss is compared with the line reference resistance loss value calculated based on the line design parameters and rated operating conditions. When the real-time resistance loss is higher than the line reference resistance loss value, the line being compared is located and marked as a high-loss line. Regional-level analysis: Calculate the regional energy efficiency index of each functional area, compare the regional energy efficiency index with the preset regional energy efficiency benchmark value, and if the regional energy efficiency index is lower than the preset regional energy efficiency benchmark value, locate the functional area being compared and mark the located area as a low-energy-efficiency area. The located low-efficiency equipment and high-loss line are determined as high-loss links in the energy flow, and the located low-energy-efficiency area is determined as a low-efficiency area in the energy flow.

[0025] It needs to be specifically pointed out that when the low-efficiency operation or the abnormally high energy consumption state is identified, the real-time current, voltage, power factor of each power branch circuit and the energy input and output data of each functional area are collected based on multi-dimensional operation data, structured energy data sets are formed through classification and integration, and a weighted directed graph energy flow network model is constructed with the input of commercial power as a source node, the power equipment as an intermediate node, and the energy dissipation form as a sink node. Through the calculation of the energy flow and loss distribution of the nodes and edges in the model, multi-level loss analysis at the device level, line level and area level is carried out, and finally the low-efficiency equipment and the high-loss line are determined as high-loss links, and the low-energy-efficiency area is determined as a low-efficiency area, so as to realize the fine positioning and classification of the energy flow loss.

[0026] S4: According to the determined high-loss links and low-efficiency areas, a multi-dimensional coupling model between devices in the machine room is constructed.

[0027] In this embodiment, the specific process of S4 for constructing a multi-dimensional coupling model according to the determined high-loss links and low-efficiency areas includes: Based on the physical topology, spatial position and operation logic between devices, the coupling relationship including energy flow and thermal dimension is defined; then a multi-dimensional coupling network model is constructed with devices as nodes and coupling relationship as edges, and the coupling strength of each edge is quantified by using historical operation data, and the determined low-efficiency equipment, high-loss line and low-efficiency area are loaded as abnormal nodes into the multi-dimensional coupling network model.

[0028] It needs to be specifically pointed out that after determining these abnormal points, a multi-dimensional coupling model is constructed. The model first defines the coupling relationship including energy flow (such as power supply relationship, energy consumption transmission) and thermal dimension (such as heat diffusion, cooling influence) based on the physical topology (such as power connection, network connection), spatial position (such as cabinet layout, cold and hot channel distribution) and operation logic (such as application dependence, control strategy) between devices, and then constructs a multi-dimensional coupling network model with devices as nodes and coupling relationship as edges, and quantifies the coupling strength of each edge by using historical operation data, and finally loads the aforementioned determined low-efficiency equipment, high-loss line and low-efficiency area as abnormal nodes into the model.

[0029] S5: Input real-time multi-dimensional operation data into the multi-dimensional coupling model, output predicted multi-dimensional operation data, generate energy-saving optimization control strategy according to the predicted multi-dimensional operation data, and execute the energy-saving optimization control strategy to adjust the device operation state.

[0030] In this embodiment, S5 generates and executes the energy-saving optimization control strategy according to the multi-dimensional coupling model, which includes: The real-time multi-dimensional running data is input into the multi-dimensional coupling network model for simulation deduction, so as to output the predicted multi-dimensional running data of the equipment running state and energy consumption response in the future time window; Based on the output predicted multi-dimensional running data and the preset energy-saving target, the optimal control parameter combination is solved by an optimization algorithm, and the optimal control parameter combination is converted into an energy-saving optimization control strategy for specific equipment, and the energy-saving optimization control strategy is executed to dynamically adjust the running state of the related equipment; The actual multi-dimensional running data after adjustment is continuously monitored, and the deviation between the actual multi-dimensional running data and the predicted multi-dimensional running data is compared, the multi-dimensional coupling network model is parameter corrected, and the actual multi-dimensional running data is used as input data and applied to the multi-dimensional coupling network model after parameter correction for the next round of optimization.

[0031] As Figure 2 The embodiment provides an anti-counterfeiting method based on local feature visual information, and an implementation system thereof is provided, which includes a multi-dimensional data acquisition module, a computer room energy consumption state recognition module, an energy flow distribution analysis module, a multi-dimensional coupling model construction module, and a control strategy generation and adjustment module. The multi-dimensional data acquisition module is connected with the computer room energy consumption state recognition module, the computer room energy consumption state recognition module is connected with the energy flow distribution analysis module, the energy flow distribution analysis module is connected with the multi-dimensional coupling model construction module, and the multi-dimensional coupling model construction module is connected with the control strategy generation and adjustment module.

[0032] The multi-dimensional data acquisition module acquires multi-dimensional running data of the computer room, and pre-processes the collected multi-dimensional running data; The computer room energy consumption state recognition module constructs a computer room energy consumption evaluation model according to the pre-processed multi-dimensional running data, evaluates the current energy consumption state of the computer room based on the computer room energy consumption evaluation model, and identifies the current energy consumption state of the computer room; The energy flow distribution analysis module analyzes the energy flow distribution in the computer room according to the identified current energy consumption state, and determines the high-loss link and low-efficiency area in the energy flow; The multi-dimensional coupling model construction module constructs a multi-dimensional coupling model between the devices in the computer room according to the determined high-loss link and low-efficiency area; The control strategy generation and adjustment module inputs real-time multi-dimensional running data into the multi-dimensional coupling model, outputs predicted multi-dimensional running data, generates an energy-saving optimization control strategy according to the predicted multi-dimensional running data, and executes the energy-saving optimization control strategy to adjust the running state of the equipment.

[0033] Finally: the above only for the preferred embodiments of the present application, and not for limiting the present application, any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application, should be included in the scope of protection of the present application.

[0034] The above description is merely a specific implementation of the present application, but the protection scope of the present application is not limited thereto, any modification, equivalent replacement, improvement, etc. made within the technical range disclosed by the present application, should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A multi-dimensional coupled energy-saving optimization control method for computer rooms, characterized in that, include: S1: Acquire multi-dimensional operational data of the computer room and preprocess the collected multi-dimensional operational data; S2: Based on the preprocessed multi-dimensional operational data, construct a data center energy consumption assessment model, assess the current energy consumption status of the data center based on the data center energy consumption assessment model, and identify the current energy consumption status of the data center. S3: Based on the identified current energy consumption status, analyze the energy flow distribution in the computer room and determine the high-loss links and low-efficiency areas in the energy flow; S4: Based on the identified high-loss links and low-efficiency areas, construct a multi-dimensional coupling model between the various devices in the computer room; S5: Input real-time multi-dimensional operating data into the multi-dimensional coupled model, output predicted multi-dimensional operating data, generate energy-saving optimization control strategy based on the predicted multi-dimensional operating data, and execute the energy-saving optimization control strategy to adjust the equipment operating status.

2. The multi-dimensional coupled data center energy-saving optimization control method according to claim 1, characterized in that, The S1 collects multi-dimensional operational data in the computer room in real time by deploying a sensor network, intelligent interfaces for IT equipment, and a power monitoring system. The multi-dimensional operational data includes ambient temperature, ambient humidity, equipment power consumption, server utilization, equipment temperature, equipment vibration frequency, network traffic, and power supply parameters. The specific operations for preprocessing the collected multi-dimensional operational data include: data alignment, data cleaning, data transformation, and data normalization.

3. The multi-dimensional coupled data center energy-saving optimization control method according to claim 2, characterized in that, S2 extracts characteristic parameters that affect the energy consumption of the computer room from the preprocessed multi-dimensional operational data by performing correlation analysis and feature extraction on the multi-dimensional operational data. Based on the extracted feature parameters, a data center energy consumption assessment model is constructed. The real-time feature parameter sequence is input into the constructed data center energy consumption assessment model, and the model outputs a theoretical optimal energy consumption value. By comparing the actual total energy consumption of the data center with the theoretical optimal energy consumption, the energy efficiency deviation index is calculated as a quantitative indicator to assess the current energy consumption status of the data center. The calculated energy efficiency deviation index is compared with multiple preset energy consumption state threshold ranges to identify and output the current energy consumption state of the computer room. The energy consumption state includes: high-efficiency operation state, normal operation state, low-efficiency operation state and abnormally high energy consumption state. If the data center is identified as operating inefficiently or with abnormally high energy consumption, the analysis process for the energy flow distribution within the data center is triggered; if the data center is identified as operating efficiently or in normal operation, the monitoring and evaluation of the data center's energy consumption status will continue.

4. The multi-dimensional coupled data center energy-saving optimization control method according to claim 3, characterized in that, When S3 detects that the computer room is in an inefficient operating state or an abnormally high energy consumption state, it immediately performs further analysis on the energy flow distribution within the computer room. The specific analysis process includes: Based on the collected multi-dimensional operational data, further collect real-time current, voltage, and power factor data of each power branch circuit, as well as energy input and output data of each functional area. Then, classify and integrate the data according to equipment type, functional area, and time dimension to form a structured energy dataset. Based on a structured energy dataset, an energy flow network model is constructed with mains power input as the source node, each electrical device as the intermediate node, and the final energy dissipation form as the sink node. The energy flow network model is a weighted directed graph, where the weight of the edge represents the magnitude of energy flow. The flow of energy and the loss distribution of energy in each node and edge of the energy flow network model are calculated. Based on the energy flow network model, a multi-level loss analysis is performed, which includes: Equipment-level analysis: This is used to analyze high-loss processes. It calculates the real-time energy utilization rate of each piece of equipment and compares it with a preset equipment efficiency benchmark value to obtain an energy utilization deviation value. When the energy utilization deviation value exceeds a preset deviation threshold, the compared equipment is identified and marked as inefficient equipment. The preset equipment efficiency benchmark value is determined based on equipment design parameters and historical best operating data, and is dynamically corrected quarterly based on equipment maintenance records and changes in operating conditions. Line-level analysis: Joule's law is used to calculate the real-time resistance loss of each power branch line, and the real-time resistance loss is compared with the line reference resistance loss value calculated based on the line design parameters and rated operating conditions. When the real-time resistance loss is higher than the line reference resistance loss value, the line being compared is located and marked as a high-loss line. Regional-level analysis: Calculate the regional energy efficiency index of each functional area, compare the regional energy efficiency index with the preset regional energy efficiency benchmark value, and if the regional energy efficiency index is lower than the preset regional energy efficiency benchmark value, locate the functional area being compared and mark the located area as a low-energy-efficiency area. The identified inefficient equipment and high-loss lines are considered as high-loss links in the energy flow, and the identified low-energy-efficiency areas are considered as low-efficiency areas in the energy flow.

5. The multi-dimensional coupled data center energy-saving optimization control method according to claim 4, characterized in that, The specific process of S4 constructing a multidimensional coupling model based on the determined high-loss links and low-efficiency regions includes: Based on the physical topology, spatial location, and operational logic between devices, a coupling relationship including energy flow and thermal dimensions is defined. Then, a multidimensional coupled network model is constructed with devices as nodes and the aforementioned coupling relationship as edges. Historical operational data is used to quantify the coupling strength of each edge, and identified inefficient devices, high-loss lines, and low-efficiency areas are loaded as abnormal nodes into the multidimensional coupled network model.

6. The multi-dimensional coupled data center energy-saving optimization control method according to claim 5, characterized in that, The energy-saving optimization control strategy generated and executed by S5 based on the multidimensional coupling model includes: Real-time multi-dimensional operating data is input into the multi-dimensional coupled network model for simulation and deduction, thereby outputting predicted multi-dimensional operating data for predicting the operating status and energy consumption response of the equipment within the future time window; Based on the output predicted multi-dimensional operating data and the preset energy-saving target, the optimal combination of control parameters is solved by the optimization algorithm. Then, the optimal combination of control parameters is transformed into an energy-saving optimization control strategy for specific equipment, and the energy-saving optimization control strategy is executed to dynamically adjust the operating status of the relevant equipment. The actual multi-dimensional operating data after adjustment is continuously monitored. By comparing the deviation between the actual multi-dimensional operating data and the predicted multi-dimensional operating data, the parameters of the multi-dimensional coupled network model are corrected. Then, the actual multi-dimensional operating data is used as input data and applied to the multi-dimensional coupled network model after parameter correction for the next round of optimization.

7. A multi-dimensional coupled data center energy-saving optimization control system, implementing the multi-dimensional coupled data center energy-saving optimization control method as described in any one of claims 1-6, characterized in that, include: Multidimensional data acquisition module: Acquires multidimensional operational data of the computer room and preprocesses the collected multidimensional operational data; Data center energy consumption status identification module: Based on preprocessed multi-dimensional operating data, a data center energy consumption assessment model is constructed. Based on the data center energy consumption assessment model, the current energy consumption status of the data center is assessed and the current energy consumption status of the data center is identified. Energy Flow Distribution Analysis Module: Based on the identified current energy consumption status, analyze the energy flow distribution within the computer room to determine high-loss links and low-efficiency areas in the energy flow; Multidimensional coupling model construction module: Based on the identified high-loss links and low-efficiency areas, construct a multidimensional coupling model between various devices in the computer room; Control strategy generation and adjustment module: Input real-time multi-dimensional operating data into the multi-dimensional coupled model, output predicted multi-dimensional operating data, generate energy-saving optimization control strategy based on the predicted multi-dimensional operating data, and execute the energy-saving optimization control strategy to adjust the equipment operating status.

Images