Micro-module machine room heat-electricity collaborative optimization method based on multi-parameter fusion control

Abstract

The invention relates to the technical field of digital twinning, and discloses a micro-module machine room heat-electricity collaborative optimization method based on multi-parameter fusion control, and the method comprises the steps: constructing a heat-electricity collaborative optimization network of a micro-module machine room; collecting multi-source heterogeneous data of the micro-module machine room in real time, and fusing the multi-source heterogeneous data to obtain fused data; and calculating a key state index of the micro-module machine room, solving a target function in the heat-electricity collaborative optimization network to obtain a collaborative optimization parameter sequence, and generating a collaborative optimization instruction of the micro-module machine room to execute heat-electricity collaborative optimization of the micro-module machine room. According to the method, the reliability and adaptability of micro-module machine room heat-electricity collaborative optimization can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to a micro-module machine room heat-electricity collaborative optimization method based on multi-parameter fusion control, belonging to the technical field of digital twinning. BACKGROUND

[0002] Micro-module machine room heat-electricity collaborative optimization refers to an advanced operation and maintenance management strategy that regards the power supply system (electricity) and the refrigeration system (heat) of the machine room as an organic whole rather than independent subsystems, and realizes the comprehensive goals of maximum energy efficiency, maximum reliability and minimum operating cost through real-time sensing, intelligent analysis and integrated control. Micro-module machine room heat-electricity collaborative optimization can significantly extend the service life of key equipment such as servers, batteries and air conditioners by precise control, and reduce capital expenditure.

[0003] Traditional micro-module machine room heat-electricity collaborative optimization is an event-triggered simple logic linkage. When the state of one system (usually refrigeration) reaches a certain critical point, the controller (usually PLC or environmental monitoring system) sends instructions to the other system (usually power supply) to adjust. This method cannot adjust itself according to real-time changes in business load, seasonal changes in outdoor climate, and equipment aging, etc. because all thresholds and strategies are pre-set and fixed, resulting in a too rigid and poor adaptive micro-module machine room heat-electricity collaborative optimization. SUMMARY

[0004] The present application provides a micro-module machine room heat-electricity collaborative optimization method and system based on multi-parameter fusion control, which aims to improve the reliability and adaptability of micro-module machine room heat-electricity collaborative optimization.

[0005] To achieve the above-mentioned purpose, the present application provides a micro-module machine room heat-electricity collaborative optimization method based on multi-parameter fusion control, comprising:

[0006] Constructing a heat-electricity collaborative optimization network of the micro-module machine room, wherein the heat-electricity collaborative optimization network comprises a multi-dimensional perception layer, an operating state analysis layer, an optimization parameter calculation layer and a collaborative optimization execution layer;

[0007] Through the multi-dimensional perception layer, multi-source heterogeneous data of the micro-module machine room is collected in real time, and the multi-source heterogeneous data is fused to obtain fusion data;

[0008] Based on the fusion data, the operating state analysis layer calculates key state indicators of the micro-module machine room, wherein the key state indicators include a heat distribution coefficient, an air flow organization effectiveness index, a refrigeration system comprehensive energy efficiency ratio and an IT load dynamic change rate;

[0009] solving the objective function in the thermal-electricity co-optimization network based on the key state indicators through the optimization parameter calculation layer to obtain a co-optimization parameter sequence, wherein the co-optimization parameters include a set temperature, a fan rotating speed, a compressor frequency, a power supply power, and an IT load migration;

[0010] generating co-optimization instructions of the micro-module machine room through the co-optimization execution layer based on the co-optimization parameter sequence to execute the thermal-electricity co-optimization of the micro-module machine room.

[0011] Optionally, the thermal-electricity co-optimization network of the micro-module machine room comprises:

[0012] collecting equipment parameters of the micro-module machine room to construct a digital layout diagram of the micro-module machine room;

[0013] defining a thermal-electricity co-optimization target of the micro-module machine room;

[0014] constructing a multi-layer network architecture of the micro-module machine room according to the thermal-electricity co-optimization target and the digital layout diagram, wherein the multi-layer network architecture comprises a multi-dimensional perception layer, an operating state analysis layer, an optimization parameter calculation layer, and a co-optimization execution layer;

[0015] determining a management network and a communication protocol between the multi-layer network architecture;

[0016] defining an operating logic between the multi-layer network architecture;

[0017] integrating the management network, the operating logic, and the communication protocol into the multi-layer network architecture to obtain a thermal-electricity co-optimization network.

[0018] Optionally, the constructing the multi-layer network architecture of the micro-module machine room according to the thermal-electricity co-optimization target comprises:

[0019] analyzing monitoring requirements of the micro-module machine room based on the thermal-electricity co-optimization target;

[0020] constructing a sensor array of the micro-module machine room based on the monitoring requirements to integrate a multi-dimensional perception layer of the micro-module machine room;

[0021] obtaining historical operating data of the micro-module machine room to construct an operating state analysis model of the micro-module machine room;

[0022] integrating an operating state analysis layer of the micro-module machine room according to the operating state analysis model;

[0023] defining an objective function and an optimization constraint condition of the micro-module machine room according to the thermal-electricity co-optimization target;

[0024] Based on the objective function and the optimization constraints, an optimization parameter calculation layer for the micro-module computer room is constructed.

[0025] Define the PLC control algorithm for the micro-module computer room to construct the collaborative optimization execution layer of the micro-module computer room.

[0026] Optionally, defining the objective function and optimization constraints of the micro-module computer room based on the thermal-electric synergistic optimization objective includes:

[0027] Based on the aforementioned thermal-electric synergistic optimization objective, the core sub-objectives of the micro-module data center are defined, wherein the core sub-objectives include: economic objective, PUE objective, cooling efficiency objective, and IT load balancing objective;

[0028] Construct sub-objective functions for each of the core sub-objectives;

[0029] The sub-objective functions are weighted and fused to obtain the objective function;

[0030] Based on the equipment parameters of the micro-module computer room, the equipment safety parameters of the micro-module computer room are determined, wherein the equipment safety parameters include: air inlet temperature parameters, equipment power consumption parameters, power supply capacity parameters, and cooling capacity parameters;

[0031] Define the equipment operation constraints of the micro-module computer room, wherein the equipment operation constraints include: air conditioning equipment constraints, load migration constraints, operation stability constraints, and operation performance constraints;

[0032] Based on the equipment operation constraints and the equipment safety parameters, the optimized constraints for the micro-module computer room are constructed.

[0033] Optionally, the fusion of the multi-source heterogeneous data to obtain fused data includes:

[0034] Identify the physical entities, logical entities, and conceptual entities of the micro-module computer room corresponding to the multi-source heterogeneous data;

[0035] Define the entity relationships between the physical entity, the logical entity, and the conceptual entity;

[0036] Based on the entity relationships, physical entities, logical entities, and conceptual entities, a knowledge graph framework for the micro-module computer room is constructed.

[0037] The multi-source heterogeneous data is preprocessed to obtain preprocessed multi-source data;

[0038] The preprocessed multi-source data is spatiotemporally aligned to obtain aligned multi-source data;

[0039] The aligned multi-source data is encapsulated using event-based methods to obtain a multi-event object;

[0040] Analyze the causal relationships between the multiple event objects and extract the multi-event features of the multiple event objects;

[0041] Based on the causal relationship and the multi-event features, the multi-event objects are mapped to the knowledge graph framework to obtain fused data.

[0042] Optionally, the step of calculating the key status indicators of the micro-module data center based on the fused data through the operational status analysis layer includes:

[0043] The fused data is input into the operation status analysis model in the operation status analysis layer, and the operation status of the micro-module computer room is output.

[0044] Extract the air intake temperature and power consumption of each server in the operating state to calculate the average temperature and temperature standard deviation of the micro-module computer room;

[0045] Calculate the heat distribution coefficient of the micro-module computer room based on the average temperature and the temperature standard deviation;

[0046] Extract the cold aisle temperature, hot aisle temperature, actual supply air temperature, and actual return air temperature from the operating state;

[0047] Based on the cold aisle temperature, the hot aisle temperature, the actual supply air temperature, and the actual return air temperature, calculate the local airflow mixing degree and the global heat recovery efficiency of the micro-module computer room;

[0048] Based on the local airflow mixing degree and the global heat recovery efficiency, the airflow organization effectiveness index of the micro-module computer room is calculated;

[0049] Extract the total instantaneous power consumption of the cooling equipment and the total instantaneous heat dissipation power of the IT equipment in the operating state to calculate the total cooling load of the micro-module computer room;

[0050] Based on the total cooling load, analyze the standard energy efficiency ratio of the micro-module computer room;

[0051] Calculate the overall energy efficiency ratio of the cooling system of the micro-module computer room based on the total cooling load and the standard energy efficiency ratio;

[0052] Extract the time series data of total IT power consumption from the fused data;

[0053] Based on the total IT power consumption time series data, calculate the dynamic change rate of IT load in the micro-module computer room;

[0054] Based on the heat distribution coefficient, the airflow organization effectiveness index, the overall energy efficiency ratio of the cooling system, and the dynamic change rate of the IT load, the key status indicators of the micro-module computer room are integrated.

[0055] Optionally, calculating the dynamic rate of change of IT load in the micro-module data center based on the total IT power consumption time series data includes:

[0056] Based on the total IT power consumption time series data, calculate the logarithmic rate of return of the micro-module data center;

[0057] Based on the total IT power consumption time series data, the fluctuation frequency and fluctuation pattern of the micro-module computer room are analyzed;

[0058] Based on the fluctuation frequency and fluctuation pattern, the volatility of the micro-module data center is analyzed;

[0059] The dynamic rate of change of IT load in the micro-module data center is calculated based on the volatility and the logarithmic return.

[0060] Optionally, based on the key state indicators, the objective function in the thermal-electric co-optimization network is solved through the optimization parameter calculation layer to obtain a co-optimization parameter sequence, including:

[0061] Based on the key status indicators, the multi-timescale analysis data of the micro-module computer room corresponding to the key status indicators is analyzed through the optimization parameter calculation layer.

[0062] Based on the multi-timescale analysis data and the objective function of the thermal-electric co-optimization network, a thermal-electric co-optimization model for the micro-module computer room is constructed.

[0063] Based on the optimization constraints of the thermal-electric synergistic optimization network, the decision variable values ​​of the micro-module computer room are calculated using the thermal-electric synergistic optimization model.

[0064] Based on the values ​​of the decision variables, a sequence of collaborative optimization parameters for the micro-module computer room is generated.

[0065] Optionally, the step of constructing the thermal-electric co-optimization model of the micro-module computer room based on the multi-timescale analysis data and the objective function of the thermal-electric co-optimization network includes:

[0066] Based on the multi-timescale analysis data, the decision variables for the micro-module data center are determined, including: set temperature, fan speed, compressor frequency, power supply, and IT load migration.

[0067] Construct a mapping relationship between the multi-timescale analysis data and the decision variables;

[0068] Based on the mapping relationship and the objective function, an initial optimization model for the micro-module computer room is constructed;

[0069] The initial optimization model is trained using a pre-set training set to obtain a trained optimization model;

[0070] Analyze the accuracy of the initial training and optimization model;

[0071] When the accuracy reaches a preset accuracy threshold, the trained and optimized model is used as the thermal-electric collaborative optimization model for the micro-module computer room.

[0072] Optionally, the step of generating collaborative optimization instructions for the micro-module data center based on the collaborative optimization parameter sequence through the collaborative optimization execution layer includes:

[0073] Determine the instruction mapping rules for the collaborative optimization execution layer;

[0074] The parameter sequence of the collaborative optimization is analyzed to obtain parameter analysis data;

[0075] According to the instruction mapping rules, the parameter parsing data is converted into initial instructions through the PLC control algorithm of the collaborative optimization execution layer;

[0076] Based on the timing requirements of the collaborative optimization parameter sequence, the execution time and execution order of the initial instructions are determined;

[0077] Based on the execution time and the execution order, the initial instructions are formatted to obtain collaborative optimization instructions.

[0078] Compared to the problems described in the background technology, this invention, by constructing the aforementioned thermal-electric synergistic optimization network, transforms the data center from a passive collection of devices into an intelligent organism with autonomous sensing, analysis, decision-making, and execution capabilities. This invention, through the fusion of multi-source heterogeneous data via a multi-dimensional sensing layer, provides the system with an unprecedented global perspective. This breaks the data isolation of various subsystems in traditional monitoring, laying a solid data foundation for accurate decision-making. Secondly, this invention, through key state indicators calculated by the operational status analysis layer, quantifies ambiguous thermal environment problems into precise numbers. This enables the system to discern the root causes of local hotspots and airflow bottlenecks, achieving an upgrade from post-event alarms to pre-event warnings and root cause analysis, greatly improving operational safety and reliability. Furthermore, the optimization parameter calculation layer is the intelligent brain of the system, no longer pursuing local optima for single systems such as cooling or power supply. Rather than simply focusing on efficiency, this invention, based on a global state, unifies and coordinates the allocation of thermal, electrical, and computing resources while meeting IT load demands. This directly leads to extreme energy efficiency, significantly reducing PUE and achieving a leap from extensive energy conservation to refined energy efficiency optimization. Finally, through a collaborative optimization execution layer, this invention transforms intelligent decisions into precise physical actions, ensuring that optimization commands can be safely and smoothly applied to fans, compressors, and even IT loads, achieving a shift from manual intervention to fully automated closed-loop control. This invention, through a closed loop of "perception-analysis-decision-execution," achieves a comprehensive leap in safety, energy efficiency, reliability, and operational efficiency. Therefore, the micro-module data center thermal-electric collaborative optimization method based on multi-parameter fusion control provided in this invention can improve the reliability and adaptability of thermal-electric collaborative optimization in micro-module data centers. Attached Figure Description

[0079] Figure 1 This is a flowchart illustrating a micro-module computer room thermal-electric synergistic optimization method based on multi-parameter fusion control, provided as an embodiment of the present invention.

[0080] Figure 2 This is a schematic diagram of a module for implementing the micro-module computer room thermal-electric synergistic optimization system based on multi-parameter fusion control, provided as an embodiment of the present invention.

[0081] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0082] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0083] This application provides a method for thermal-electrical coordinated optimization of micro-module data centers based on multi-parameter fusion control. The executing entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the method can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster.

[0084] Reference Figure 1 The diagram shown is a flowchart illustrating a micro-module data center thermal-electrical co-optimization method based on multi-parameter fusion control according to an embodiment of the present invention. In this embodiment, the micro-module data center thermal-electrical co-optimization method based on multi-parameter fusion control includes:

[0085] S1. Construct a thermal-electric collaborative optimization network for the micro-module computer room, wherein the thermal-electric collaborative optimization network includes a multi-dimensional perception layer, an operation status analysis layer, an optimization parameter calculation layer, and a collaborative optimization execution layer.

[0086] This invention, through the construction of a thermal-electricity coordinated optimization network for micro-modular data centers, can proactively place some servers into deep sleep mode while simultaneously increasing air conditioning set temperatures and reducing fan speeds. This enhances cooling in areas where needed, avoiding large-scale energy waste and significantly improving energy efficiency and operational safety. The thermal-electricity coordinated optimization network refers to an intelligent network architecture specifically designed for micro-modular data centers, based on data-driven and intelligent decision-making, integrating sensing, analysis, decision-making, and execution capabilities.

[0087] As an embodiment of the present invention, the construction of the thermal-electric collaborative optimization network for the micro-module computer room includes:

[0088] Collect the equipment parameters of the micro-module computer room to construct a digital layout diagram of the module computer room;

[0089] Define the thermal-electric synergistic optimization objectives for the micro-module computer room;

[0090] Based on the thermal-electric synergistic optimization objective and the digital layout diagram, a multi-layer network architecture for the micro-module computer room is constructed, wherein the multi-layer network architecture includes: a multi-dimensional perception layer, an operation status analysis layer, an optimization parameter calculation layer, and a collaborative optimization execution layer;

[0091] Determine the management network and communication protocols between the multi-layered network architecture;

[0092] Define the operational logic between the aforementioned multi-layer network architecture;

[0093] The management network, the operating logic, and the communication protocol are integrated into the multi-layer network architecture to obtain a thermal-electric synergistic optimization network.

[0094] The equipment parameters refer to the specific specifications and operational information of all hardware devices constituting the micro-module data center, such as the model, power, location, and interface type of servers, cabinets, air conditioners, and power distribution units. The digital layout diagram refers to a precise, visualized two-dimensional model of the micro-module data center. The thermal-electric synergistic optimization objective aims to simultaneously optimize the data center's energy efficiency, cooling effect, IT equipment operational security, and operating costs, achieving a dynamic balance between heat generation and dissipation, and power supply and consumption. The multi-layer network architecture refers to a layered technical framework designed to achieve thermal-electric synergistic optimization. The multi-dimensional perception layer is a network layer used to acquire real-time equipment operating status and environmental parameters. The operational status analysis layer is a network layer used to receive data from the perception layer and use an analysis model to evaluate the current data center operating status. The optimization parameter calculation layer is a network layer that calculates the specific control parameters to achieve the optimal state by solving the objective function. The synergistic optimization execution layer is a network layer that converts the optimization parameters output by the calculation layer into specific control instructions and directly adjusts the operation of equipment such as air conditioners and power supplies through a programmable logic controller (PLC). The management network refers to a dedicated data communication network used to transmit control commands, configuration information, and status feedback, ensuring reliable and secure communication between various parts of the optimization system. The communication protocol refers to the rules and format standards followed when transmitting data on the management network, such as MQTT, Modbus, and OPC UA. The operational logic refers to the rules and processes that define how data flows from the perception layer to the analysis layer, how analysis results trigger optimization tasks in the computation layer, and how computation results are applied by the execution layer, forming a closed-loop automated optimization process.

[0095] Optionally, the digital layout diagram of the modular computer room can be constructed using deep learning models, such as YOLO and CNN.

[0096] Optionally, the management network between the multi-layer network architectures can be determined by an intent network, such as defining management intents between the multi-layer network architectures and converting the management intents into network configurations through the intent network to determine the management network between the multi-layer network architectures.

[0097] Optionally, the operational logic between the multi-layer network architecture can be defined by the causal reasoning logic of the knowledge graph, such as constructing a knowledge graph of the multi-layer network architecture to analyze the relationships between the layers of the multi-layer network architecture, and defining the operational logic between the multi-layer network architecture based on the relationships between the layers.

[0098] Optionally, constructing the multi-layer network architecture of the micro-module computer room based on the thermal-electric synergistic optimization objective includes:

[0099] Based on the aforementioned thermal-electric synergistic optimization objective, the monitoring requirements of the micro-module computer room are analyzed;

[0100] Based on the aforementioned monitoring requirements, a sensor array for the micro-module computer room is constructed to integrate the multi-dimensional sensing layer of the micro-module computer room.

[0101] Historical operational data of the micro-module data center is obtained to construct an operational status analysis model for the micro-module data center;

[0102] Based on the aforementioned operational status analysis model, the operational status analysis layer of the micro-module computer room is integrated;

[0103] Based on the aforementioned thermal-electric synergistic optimization objective, the objective function and optimization constraints of the micro-module computer room are defined.

[0104] Based on the objective function and the optimization constraints, an optimization parameter calculation layer for the micro-module computer room is constructed.

[0105] Define the PLC control algorithm for the micro-module computer room to construct the collaborative optimization execution layer of the micro-module computer room.

[0106] The monitoring requirements refer to the specific requirements for the types, accuracy, location, and frequency of data to be collected to achieve the thermal-electric synergistic optimization goal. For example, monitoring server intake air temperature, air conditioner return air temperature, and real-time power consumption of server racks is required. The sensor array refers to the physical layout and logical set of sensors deployed within the micro-module server room according to the monitoring requirements. The historical operating data refers to the operating records accumulated by the micro-module server room over a period of time, including equipment power consumption, temperature changes, and cooling system operating status, serving as the data foundation for building and training the operating status analysis model. The operating status analysis model is a mathematical model used to accurately determine and predict the current and future thermal and electrical operating status of the server room based on real-time input sensing data. The objective function is a mathematical expression used to quantify the optimization effect. The optimization constraints refer to a series of restrictions that must be followed during the optimization process, such as equipment power not exceeding rated values ​​and server intake air temperature being within a safe range. The PLC control algorithm refers to the control algorithm written in a programmable logic controller (PLC) for directly controlling the underlying hardware program logic.

[0107] Optionally, the monitoring requirements of the micro-module computer room can be analyzed using causal inference techniques, such as constructing a variable causal graph of the micro-module computer room, identifying key causal paths in the variable causal graph based on the thermal-electric synergistic optimization objective, and thus determining the monitoring requirements of the micro-module computer room.

[0108] Optionally, the operational state analysis model can be constructed using convolutional networks, such as spatiotemporal graph convolutional networks.

[0109] Optionally, defining the objective function and optimization constraints of the micro-module computer room based on the thermal-electric synergistic optimization objective includes:

[0110] Based on the aforementioned thermal-electric synergistic optimization objective, the core sub-objectives of the micro-module data center are defined, wherein the core sub-objectives include: economic objective, PUE objective, cooling efficiency objective, and IT load balancing objective;

[0111] Construct sub-objective functions for each of the core sub-objectives;

[0112] The sub-objective functions are weighted and fused to obtain the objective function;

[0113] Based on the equipment parameters of the micro-module computer room, the equipment safety parameters of the micro-module computer room are determined, wherein the equipment safety parameters include: air inlet temperature parameters, equipment power consumption parameters, power supply capacity parameters, and cooling capacity parameters;

[0114] Define the equipment operation constraints of the micro-module computer room, wherein the equipment operation constraints include: air conditioning equipment constraints, load migration constraints, operation stability constraints, and operation performance constraints;

[0115] Based on the equipment operation constraints and the equipment safety parameters, the optimized constraints for the micro-module computer room are constructed.

[0116] The core sub-objectives refer to several key and independent optimization directions that constitute the overall thermal-electric synergistic optimization objective, including economy, energy efficiency, cooling effect, and load distribution. The economic objective aims to minimize the operating costs of the data center, primarily including electricity costs (IT equipment electricity costs and cooling electricity costs). The PUE objective aims to minimize the Power Usage Effectiveness (PUE) value, i.e., reducing the ratio of total energy consumption to IT equipment energy consumption. The cooling efficiency objective aims to remove heat generated by IT equipment in the data center accurately and efficiently with minimal energy consumption, ensuring all equipment operates within a safe temperature range. The IT load balancing objective aims to achieve a more balanced and rational utilization of computing, power, and cooling resources in the data center by intelligently scheduling and migrating IT loads (such as virtual machines and containers) while meeting business performance requirements, thereby supporting improvements in overall energy efficiency and security. Equipment safety parameters refer to the limits specified by the equipment manufacturer to ensure the safe and stable operation of the equipment, such as the maximum allowable inlet air temperature and maximum power consumption. The inlet air temperature parameter refers to the highest and lowest allowable temperatures at the fan inlet of the IT equipment. The equipment power consumption parameter refers to the maximum allowable power consumption value for a single device. The power supply capacity parameter refers to the maximum power capacity that the power distribution unit can provide. The cooling capacity parameter refers to the maximum cooling capacity that the air conditioning system can provide. The equipment operation constraints refer to the operational limitations set during the optimization process to ensure the stable, safe, and efficient operation of the system. The air conditioning equipment constraints refer to the limitations set on the physical characteristics and operating rules of the air conditioning equipment itself. The load migration constraints refer to the limitations set when migrating IT workloads between different physical servers. The operational stability constraints refer to the limitations set to prevent the optimization system from making overly drastic, frequent, or oscillating adjustments. The operational performance constraints refer to the minimum requirements set to ensure the quality of IT business services.

[0117] Optionally, the sub-objective function of the core sub-objective can be constructed using a machine learning model, such as a gradient boosting tree or a neural network.

[0118] Optionally, the equipment operation constraints of the micro-module computer room can be defined using causal inference techniques, such as Granger causality test, Do-Calculus, etc.

[0119] S2. Through the multi-dimensional perception layer, multi-source heterogeneous data of the micro-module computer room is collected in real time, and the multi-source heterogeneous data is fused to obtain fused data.

[0120] This invention, through its multi-dimensional perception layer, effectively breaks down data silos by collecting multi-source heterogeneous data from the micro-module data center in real time, providing comprehensive real-time data for subsequent analysis and optimization. The multi-source heterogeneous data refers to a comprehensive set of data reflecting the operational status of the data center, collected in real time from various devices and sensors of different sources, types, and formats within the micro-module data center, such as server power consumption, temperature sensor data, network traffic, and air conditioning parameters.

[0121] This invention, through the fusion of multi-source heterogeneous data, yields fused data that can identify and eliminate outliers, noise, and missing values ​​from individual data sources. This improves the accuracy, consistency, and completeness of the data, enhances the reliability of subsequent analysis results, and provides a comprehensive, multi-dimensional data view reflecting the operational status of the micro-module data center, offering a complete data foundation for systematic analysis and decision-making. The fused data refers to a structured, high-dimensional dataset with inherent logical connections, formed after a series of standardization, alignment, correlation, cleaning, and enhancement processes from the original multi-source heterogeneous data collected from the multi-dimensional perception layer.

[0122] As an embodiment of the present invention, the fusion of the multi-source heterogeneous data to obtain fused data includes:

[0123] Identify the physical entities, logical entities, and conceptual entities of the micro-module computer room corresponding to the multi-source heterogeneous data;

[0124] Define the entity relationships between the physical entity, the logical entity, and the conceptual entity;

[0125] Based on the entity relationships, physical entities, logical entities, and conceptual entities, a knowledge graph framework for the micro-module computer room is constructed.

[0126] The multi-source heterogeneous data is preprocessed to obtain preprocessed multi-source data;

[0127] The preprocessed multi-source data is spatiotemporally aligned to obtain aligned multi-source data;

[0128] The aligned multi-source data is encapsulated using event-based methods to obtain a multi-event object;

[0129] Analyze the causal relationships between the multiple event objects and extract the multi-event features of the multiple event objects;

[0130] Based on the causal relationship and the multi-event features, the multi-event objects are mapped to the knowledge graph framework to obtain fused data.

[0131] The physical entities refer to independent devices, components, and space units that objectively exist in the physical space of the micro-module data center and can be directly perceived or located, such as servers, racks, precision air conditioners, UPS, PDUs, etc. The logical entities refer to abstract units composed of one or more physical entities through software configuration, network protocols, or virtualization technology, existing as a whole in function or management, such as virtual machines, containers, VLANs (Virtual Local Area Networks), application clusters, etc. The conceptual entities refer to non-material concepts used to describe the state, performance, constraints, or abstract classification of physical and logical entities, such as Power Usage Effectiveness (PUE), alarm levels, risk thresholds, and failure modes, etc. The entity relationships refer to semantic links defining the connections and interactions between physical, logical, and conceptual entities, such as server A located in rack B, virtual machine C running on server A, and air conditioner E cooling server D, etc. The knowledge graph framework refers to a predefined, structured model used to describe knowledge in the micro-module data center domain. The preprocessed multi-source data refers to data that has undergone cleaning, format conversion, outlier handling, and missing value imputation to become regular, consistent, and usable for subsequent analysis. The aligned multi-source data refers to a unified and collaborative data view formed by precisely matching and associating data from different data sources in the time and spatial dimensions through spatiotemporal alignment technology after preprocessing. The multi-event object refers to a structured data unit with clear semantics describing a state change, encapsulated according to time windows and business logic after the aligned multi-source data has been processed. The causal relationship refers to the logical connection between different event objects, where the occurrence of one event directly or indirectly leads to the occurrence of another event. The multi-event feature refers to the key attributes extracted from single or multiple event objects that can be used to describe, distinguish, and quantify the essential characteristics of the event. For example, for a server temperature anomaly event, its features may include: temperature peak, duration, temperature change rate, associated average CPU utilization, and the average temperature of the server rack.

[0132] Optionally, the entity relationships between the physical entity, the logical entity, and the conceptual entity can be defined using an attribute graph model, such as Neo4j.

[0133] Optionally, the causal relationship between the multiple event objects can be analyzed by causal inference algorithms, such as the PC algorithm, the FCI algorithm, or methods based on Granger causality tests.

[0134] Optionally, the fused data can be obtained through semantically enhanced event mapping techniques, such as using natural language processing (NLP) and semantic analysis techniques to convert event descriptions into structured semantic representations, and then matching them with entities and entity relationships in a knowledge graph framework to perform data fusion and obtain fused data.

[0135] S3. Based on the fused data, the key status indicators of the micro-module computer room are calculated through the operation status analysis layer. The key status indicators include: heat distribution coefficient, airflow organization effectiveness index, cooling system comprehensive energy efficiency ratio, and IT load dynamic change rate.

[0136] This invention, through its embodiment, calculates key status indicators for the micro-module data center based on the fused data via the operational status analysis layer. This transforms the fused data into a direct and quantitative assessment of the micro-module data center's health, efficiency, and risk, providing a solid data foundation and clear optimization directions for subsequent decision-making and control. The key status indicators refer to a set of high-level, refined, and physically meaningful metrics used to quantitatively evaluate the micro-module data center's operational performance, health level, and efficiency across specific dimensions.

[0137] As an embodiment of the present invention, the calculation of key status indicators of the micro-module data center based on the fused data through the operation status analysis layer includes:

[0138] The fused data is input into the operation status analysis model in the operation status analysis layer, and the operation status of the micro-module computer room is output.

[0139] Extract the air intake temperature and power consumption of each server in the operating state to calculate the average temperature and temperature standard deviation of the micro-module computer room;

[0140] Calculate the heat distribution coefficient of the micro-module computer room based on the average temperature and the temperature standard deviation;

[0141] Extract the cold aisle temperature, hot aisle temperature, actual supply air temperature, and actual return air temperature from the operating state;

[0142] Based on the cold aisle temperature, the hot aisle temperature, the actual supply air temperature, and the actual return air temperature, calculate the local airflow mixing degree and the global heat recovery efficiency of the micro-module computer room;

[0143] Based on the local airflow mixing degree and the global heat recovery efficiency, the airflow organization effectiveness index of the micro-module computer room is calculated;

[0144] Extract the total instantaneous power consumption of the cooling equipment and the total instantaneous heat dissipation power of the IT equipment in the operating state to calculate the total cooling load of the micro-module computer room;

[0145] Based on the total cooling load, analyze the standard energy efficiency ratio of the micro-module computer room;

[0146] Calculate the overall energy efficiency ratio of the cooling system of the micro-module computer room based on the total cooling load and the standard energy efficiency ratio;

[0147] Extract the time series data of total IT power consumption from the fused data;

[0148] Based on the total IT power consumption time series data, calculate the dynamic change rate of IT load in the micro-module computer room;

[0149] Based on the heat distribution coefficient, the airflow organization effectiveness index, the overall energy efficiency ratio of the cooling system, and the dynamic change rate of the IT load, the key status indicators of the micro-module computer room are integrated.

[0150] The "operational status" refers to a snapshot-style, high-fidelity digital description of the state of all physical and logical entities within the micro-module data center, generated by the operational status analysis model based on fused data at a specific point in time. The "intake air temperature" refers to the temperature of the cold air drawn in from the front of the server chassis. The "server power consumption" refers to the electrical energy consumed by the server at the current moment. The "average temperature" refers to the power consumption-weighted average intake air temperature. The "temperature standard deviation" is a statistical measure of the dispersion of the intake air temperature of all servers from their power consumption-weighted average temperature. The "heat distribution coefficient" is a single numerical quantitative evaluation of the thermal environment health status of the micro-module data center, combining the average temperature and temperature standard deviation. The "cold aisle temperature" refers to the average temperature measured by multiple sensors within the cold aisle. The "hot aisle temperature" refers to the average temperature measured by multiple sensors within the hot aisle. The "actual supply air temperature" refers to the average temperature at the air outlet of the air conditioning unit. The "actual return air temperature" refers to the average temperature at the return air inlet of the air conditioning unit. The "local airflow mixing degree" is an indicator measuring the degree of mixing of cold and hot air in a designated area within the data center. The global heat recovery efficiency refers to a macroscopic indicator measuring the ability of a data center cooling system to recover and reuse energy from exhausted hot air. The airflow organization effectiveness index is an indicator used to evaluate the overall quality of the airflow organization design of the entire micro-module server room. The total instantaneous power consumption refers to the sum of the power consumption of all cooling-related equipment such as air conditioning compressors, fans, and water pumps at the current moment. The total instantaneous heat dissipation power refers to the sum of the power consumption of all IT equipment such as servers and network devices at the current moment. The total cooling load refers to the total heat that the data center cooling system must remove from IT equipment and its operating environment per unit time. The comprehensive energy efficiency ratio of the cooling system is a macroscopic indicator measuring the overall energy efficiency of the data center cooling system. The IT total power consumption time series data refers to a series of IT total power consumption values ​​arranged in chronological order. The IT load dynamic change rate is an indicator measuring the degree of fluctuation in the total power consumption of data center IT equipment over time.

[0151] Optionally, the average temperature and temperature standard deviation of the micro-module computer room can be calculated using the power consumption weighted average method, and the temperature standard deviation can be calculated using the weighted standard deviation formula.

[0152] Optionally, the total cooling load of the micro-module computer room can be calculated using machine learning algorithms, such as gradient boosting trees, long short-term memory networks (LSTM), etc.

[0153] As another implementation, the local airflow mixing degree and the global heat recovery efficiency are calculated using the following formulas:

[0154]

[0155]

[0156] in, Indicates the first The local airflow mixing degree of each cold aisle Indicates the global heat recovery efficiency. Indicates the first The cold aisle temperature (unit: K). Indicates the actual supply air temperature (unit: K). Indicates the number of hot aisles. Indicates the first The temperature of each thermal aisle (unit: K). This indicates the actual return air temperature (unit: K).

[0157] As another implementation, the airflow organization effectiveness index is calculated using the following formula:

[0158]

[0159] in, Indicating the airflow organization effectiveness index, Indicates the first The local airflow mixing degree of each cold aisle Indicates the global heat recovery efficiency. The weight representing the average local airflow mixing degree. The weight representing the global heat recovery efficiency.

[0160] It needs to be explained that in this application, the formula... The weight representing the average local airflow mixing degree, with a value range of (0,1). The weight representing the global heat recovery efficiency ranges from (0,1) and satisfies the following conditions: .

[0161] Optionally, calculating the dynamic rate of change of IT load in the micro-module data center based on the total IT power consumption time series data includes:

[0162] Based on the total IT power consumption time series data, calculate the logarithmic rate of return of the micro-module data center;

[0163] Based on the total IT power consumption time series data, the fluctuation frequency and fluctuation pattern of the micro-module computer room are analyzed;

[0164] Based on the fluctuation frequency and fluctuation pattern, the volatility of the micro-module data center is analyzed;

[0165] The dynamic rate of change of IT load in the micro-module data center is calculated based on the volatility and the logarithmic return.

[0166] The logarithmic rate of return refers to an indicator that measures the relative rate of change of total IT power consumption between two consecutive points in time. The fluctuation frequency refers to the number of times total IT power consumption fluctuates significantly per unit time. The fluctuation pattern refers to the inherent structure, pattern, and distribution characteristics of the fluctuations in total IT power consumption. The volatility is a quantitative measure of the drastic change in total IT power consumption.

[0167] Optionally, the fluctuation frequency of the micro-module computer room can be analyzed by Fourier transform, and the fluctuation pattern can be analyzed by visualization analysis methods, such as time series diagrams, histograms, QQ diagrams, etc.

[0168] As another implementation, the logarithmic rate of return can be calculated using the following formula:

[0169]

[0170] in, express Logarithmic return at time step Let e ​​represent the logarithmic function with base e. express Time series data of total IT power consumption at any given moment. express Time series data of total IT power consumption at any given moment.

[0171] As another implementation, the IT load dynamic change rate is calculated using the following formula:

[0172]

[0173] in, Indicates the dynamic rate of change of IT load. Indicates the attenuation factor. express Volatility at any given moment express Logarithmic return at time step.

[0174] It needs to be explained that in this application, the formula... This represents the attenuation factor, with a value range of [0.94, 0.99]. If the IT load in the micro-module data center is very regular and doesn't change much, only showing a slow growth or decline trend, then... The value is 0.99; if the IT load of the micro-module data center is elastic, it will frequently and significantly scale up and down according to business needs, then The value is 0.94.

[0175] S4. Based on the key state indicators, the objective function in the thermal-electric co-optimization network is solved through the optimization parameter calculation layer to obtain the co-optimization parameter sequence, wherein the co-optimization parameters include set temperature, fan speed, compressor frequency, power supply and IT load migration.

[0176] This invention, through the optimization parameter calculation layer based on the key state indicators, solves the objective function in the thermal-electrical collaborative optimization network. The resulting collaborative optimization parameter sequence is no longer based on fixed, preset rules. Instead, it is fine-tuned according to real-time key state indicators, achieving dynamic load balancing and hotspot elimination. This distributes heat evenly throughout the data center, avoiding the risk of equipment throttling, downtime, or even damage due to localized overheating, thereby improving the overall system reliability. The collaborative optimization parameter sequence refers to a set of parameters arranged chronologically, calculated through collaborative optimization, used to precisely regulate the operating status of data center IT equipment and cooling equipment.

[0177] As an embodiment of the present invention, the objective function in the thermal-electric co-optimization network is solved based on the key state index through the optimization parameter calculation layer to obtain the co-optimization parameter sequence, including:

[0178] Based on the key status indicators, the multi-timescale analysis data of the micro-module computer room corresponding to the key status indicators is analyzed through the optimization parameter calculation layer.

[0179] Based on the multi-timescale analysis data and the objective function of the thermal-electric co-optimization network, a thermal-electric co-optimization model for the micro-module computer room is constructed.

[0180] Based on the optimization constraints of the thermal-electric synergistic optimization network, the decision variable values ​​of the micro-module computer room are calculated using the thermal-electric synergistic optimization model.

[0181] Based on the values ​​of the decision variables, a sequence of collaborative optimization parameters for the micro-module computer room is generated.

[0182] The multi-timescale analysis data refers to a structured, hierarchical dataset formed by statistically analyzing and predicting the trends of key status indicators of the micro-module data center across different time dimensions. The thermal-electric synergistic optimization model is a mathematical model used to accurately describe the energy transfer relationship between heat generation from IT equipment and heat dissipation from cooling equipment within the micro-module data center. The decision variable values ​​are the specific numerical values ​​calculated for each decision variable after solving the aforementioned thermal-electric synergistic optimization model.

[0183] Optionally, the multi-timescale analysis data can be analyzed using deep learning techniques, such as Transformer, Informer, and Autoformer.

[0184] Optionally, the collaborative optimization parameter sequence can be generated by dynamic optimization methods, such as MPC, rolling time-domain estimation, etc.

[0185] Optionally, the step of constructing the thermal-electric co-optimization model of the micro-module computer room based on the multi-timescale analysis data and the objective function of the thermal-electric co-optimization network includes:

[0186] Based on the multi-timescale analysis data, the decision variables for the micro-module data center are determined, including: set temperature, fan speed, compressor frequency, power supply, and IT load migration.

[0187] Construct a mapping relationship between the multi-timescale analysis data and the decision variables;

[0188] Based on the mapping relationship and the objective function, an initial optimization model for the micro-module computer room is constructed;

[0189] The initial optimization model is trained using a pre-set training set to obtain a trained optimization model;

[0190] Analyze the accuracy of the initial training and optimization model;

[0191] When the accuracy reaches a preset accuracy threshold, the trained and optimized model is used as the thermal-electric collaborative optimization model for the micro-module computer room.

[0192] In this context, the decision variables refer to the variable parameters that the micro-module data center can actively control and adjust during the optimization process to achieve the optimization goal. The set temperature refers to the target outlet air temperature of the data center cooling system. The fan speed refers to the rotational speed of the precision air conditioner. The compressor frequency refers to the operating frequency of the variable frequency compressor. The power supply refers to the total power supplied to the entire micro-module data center at a specific point in time or within a specific time period. The IT load migration refers to the process of moving running computing tasks from one physical server to another in real time and seamlessly. The mapping relationship refers to the logical connection between decision variables in multi-timescale analysis data outputs, such as linear or nonlinear relationships. The initial optimization model refers to the prototype built based on theory, prior knowledge, and model structure before training and optimization. The preset training set refers to a large set of historical data samples used to train the initial optimization model, enabling it to learn the correct mapping relationship. The training optimization model refers to the new model where, after training the initial optimization model using the preset training set, it continuously adjusts its internal parameters through a learning algorithm, thereby mastering the mapping relationship between input and output. The preset accuracy threshold refers to a pre-set minimum performance standard used to determine whether the trained and optimized model is qualified.

[0193] Optionally, the mapping relationship between the multi-timescale analysis data and the decision variables can be analyzed using causal inference techniques.

[0194] Optionally, the accuracy of the initial training optimization model can be analyzed using statistical error metrics.

[0195] S5. Based on the collaborative optimization parameter sequence, a collaborative optimization instruction for the micro-module computer room is generated through the collaborative optimization execution layer to execute the thermal-electric collaborative optimization of the micro-module computer room.

[0196] This invention, through a collaborative optimization execution layer based on the aforementioned collaborative optimization parameter sequence, generates collaborative optimization instructions for the micro-module data center. This avoids instantaneous hotspots caused by sudden increases in IT load, eliminates response delays in the cooling system, and achieves synchronization among IT, thermal, and electrical systems, transforming passive response into proactive prevention. Specifically, the collaborative optimization instructions refer to a series of standardized, structured control commands generated by the collaborative optimization execution layer from the collaborative optimization parameter sequence generated by the optimization parameter calculation layer. These commands are directed towards the three domains of the micro-module data center—cooling system, power supply and distribution system, and IT system—and have clear objectives, execution paths, and timing requirements, allowing for direct parsing and execution by equipment.

[0197] As an embodiment of the present invention, the step of generating collaborative optimization instructions for the micro-module data center through a collaborative optimization execution layer based on the collaborative optimization parameter sequence includes:

[0198] Determine the instruction mapping rules for the collaborative optimization execution layer;

[0199] The parameter sequence of the collaborative optimization is analyzed to obtain parameter analysis data;

[0200] According to the instruction mapping rules, the parameter parsing data is converted into initial instructions through the PLC control algorithm of the collaborative optimization execution layer;

[0201] Based on the timing requirements of the collaborative optimization parameter sequence, the execution time and execution order of the initial instructions are determined;

[0202] Based on the execution time and the execution order, the initial instructions are formatted to obtain collaborative optimization instructions.

[0203] The instruction mapping rule refers to a predefined logical dictionary and conversion protocol used to convert abstract optimization parameters into specific device-executable actions. The parameter parsing data refers to a set of structured, machine-readable intermediate data formed after the co-optimized parameter sequence has been formatted, decomposed, and preliminarily verified. The initial instruction refers to the original control command, specific to a particular device but without considering execution timing, converted from each task item in the parameter parsing data according to the instruction mapping rule. The timing requirement refers to the time constraints and logical relationships regarding instruction execution inherent in the co-optimized parameter sequence. The execution time refers to a specific and precise execution time point calculated for each initial instruction according to the timing requirements. The execution order refers to an execution queue obtained by sorting all initial instructions according to the logical dependencies in the timing requirements.

[0204] Optionally, the parameter parsing data can be obtained through large language models, such as GPT-4, CodeLlama, etc.

[0205] Optionally, the execution time and execution order of the initial instructions can be determined by metaheuristic algorithms, such as genetic algorithms, ant colony algorithms, simulated annealing, etc.

[0206] The embodiments of the present invention can break down the barriers between thermal and electrical systems by performing thermal-electric synergistic optimization of the micro-module computer room, creating a more ideal operating environment for IT equipment, thereby reducing the energy consumption of the IT equipment itself, avoiding frequent start-ups and shutdowns and high-power operation of the cooling system, and improving the comfort of the micro-module computer room.

[0207] Compared to the problems described in the background technology, this invention, by constructing the aforementioned thermal-electric synergistic optimization network, transforms the data center from a passive collection of devices into an intelligent organism with autonomous sensing, analysis, decision-making, and execution capabilities. This invention, through the fusion of multi-source heterogeneous data via a multi-dimensional sensing layer, provides the system with an unprecedented global perspective. This breaks the data isolation of various subsystems in traditional monitoring, laying a solid data foundation for accurate decision-making. Secondly, this invention, through key state indicators calculated by the operational status analysis layer, quantifies ambiguous thermal environment problems into precise numbers. This enables the system to discern the root causes of local hotspots and airflow bottlenecks, achieving an upgrade from post-event alarms to pre-event warnings and root cause analysis, greatly improving operational safety and reliability. Furthermore, the optimization parameter calculation layer is the intelligent brain of the system, no longer pursuing local optima for single systems such as cooling or power supply. Rather than simply focusing on efficiency, this invention, based on a global state, unifies and coordinates the allocation of thermal, electrical, and computing resources while meeting IT load demands. This directly leads to extreme energy efficiency, significantly reducing PUE and achieving a leap from extensive energy conservation to refined energy efficiency optimization. Finally, through a collaborative optimization execution layer, this invention transforms intelligent decisions into precise physical actions, ensuring that optimization commands can be safely and smoothly applied to fans, compressors, and even IT loads, achieving a shift from manual intervention to fully automated closed-loop control. This invention, through a closed loop of "perception-analysis-decision-execution," achieves a comprehensive leap in safety, energy efficiency, reliability, and operational efficiency. Therefore, the micro-module data center thermal-electric collaborative optimization method based on multi-parameter fusion control provided in this invention can improve the reliability and adaptability of thermal-electric collaborative optimization in micro-module data centers.

[0208] like Figure 2 The diagram shown is a functional block diagram of a micro-module computer room thermal-electric synergistic optimization system based on multi-parameter fusion control according to the present invention.

[0209] The micro-module data center thermal-electric co-optimization system 200 based on multi-parameter fusion control described in this invention can be installed in electronic devices. According to the functions implemented, the micro-module data center thermal-electric co-optimization system includes an optimization network construction module 201, a multi-dimensional sensing module 202, a key state indicator analysis module 203, an optimization parameter sequence generation module 204, and a co-optimization execution module 205. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, stored in the memory of the electronic device.

[0210] In this embodiment of the invention, the functions of each module / unit are as follows:

[0211] The optimized network construction module 201 is used to construct a thermal-electric collaborative optimization network for the micro-module computer room, wherein the thermal-electric collaborative optimization network includes a multi-dimensional perception layer, an operation status analysis layer, an optimization parameter calculation layer, and a collaborative optimization execution layer;

[0212] The multi-dimensional perception module 202 is used to collect multi-source heterogeneous data of the micro-module computer room in real time through the multi-dimensional perception layer, and fuse the multi-source heterogeneous data to obtain fused data;

[0213] The key status indicator analysis module 203 is used to calculate the key status indicators of the micro-module computer room based on the fused data and through the operation status analysis layer. The key status indicators include: heat distribution coefficient, airflow organization effectiveness index, cooling system comprehensive energy efficiency ratio, and IT load dynamic change rate.

[0214] The optimization parameter sequence generation module 204 is used to solve the objective function in the thermal-electric co-optimization network based on the key state indicators and through the optimization parameter calculation layer to obtain the co-optimization parameter sequence. The co-optimization parameters include set temperature, fan speed, compressor frequency, power supply and IT load migration.

[0215] The collaborative optimization execution module 205 is used to generate collaborative optimization instructions for the micro-module computer room based on the collaborative optimization parameter sequence through the collaborative optimization execution layer, so as to execute the thermal-electric collaborative optimization of the micro-module computer room.

[0216] In detail, the modules in the micro-module computer room thermal-electric collaborative optimization system 200 based on multi-parameter fusion control described in this embodiment of the invention adopt the same approach as described above when in use. Figure 1 The method described herein uses the same technical means as the micro-module computer room thermal-electric synergistic optimization method based on multi-parameter fusion control, and can produce the same technical effect, so it will not be elaborated here.

[0217] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0218] Finally, it should be noted that in the above embodiments, each embodiment can be combined with each other or independent. Deleting any one of them will not affect the technical implementation of other embodiments. The above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A micro-module computer room thermal-electrical synergistic optimization method based on multi-parameter fusion control, characterized in that, The method includes: A thermal-electric collaborative optimization network for a micro-module computer room is constructed, wherein the thermal-electric collaborative optimization network includes a multi-dimensional perception layer, an operation status analysis layer, an optimization parameter calculation layer, and a collaborative optimization execution layer; Through the multi-dimensional perception layer, multi-source heterogeneous data of the micro-module computer room is collected in real time, and the multi-source heterogeneous data is fused to obtain fused data; Based on the fused data, the key status indicators of the micro-module computer room are calculated through the operation status analysis layer. The key status indicators include: heat distribution coefficient, airflow organization effectiveness index, cooling system comprehensive energy efficiency ratio, and IT load dynamic change rate. Based on the key state indicators, the objective function in the thermal-electric co-optimization network is solved through the optimization parameter calculation layer to obtain the co-optimization parameter sequence, wherein the co-optimization parameters include set temperature, fan speed, compressor frequency, power supply and IT load migration. Based on the aforementioned collaborative optimization parameter sequence, collaborative optimization instructions for the micro-module data center are generated through the collaborative optimization execution layer to execute the thermal-electric collaborative optimization of the micro-module data center.

2. The micro-module computer room thermal-electric synergistic optimization method based on multi-parameter fusion control as described in claim 1, characterized in that, The construction of the thermal-electric collaborative optimization network for the micro-module computer room includes: Collect the equipment parameters of the micro-module computer room to construct a digital layout diagram of the module computer room; Define the thermal-electric synergistic optimization objectives for the micro-module computer room; Based on the thermal-electric synergistic optimization objective and the digital layout diagram, a multi-layer network architecture for the micro-module computer room is constructed, wherein the multi-layer network architecture includes: a multi-dimensional perception layer, an operation status analysis layer, an optimization parameter calculation layer, and a collaborative optimization execution layer; Determine the management network and communication protocols between the multi-layered network architecture; Define the operational logic between the aforementioned multi-layer network architecture; The management network, the operating logic, and the communication protocol are integrated into the multi-layer network architecture to obtain a thermal-electric synergistic optimization network.

3. The micro-module computer room thermal-electric synergistic optimization method based on multi-parameter fusion control as described in claim 2, characterized in that, The construction of the multi-layer network architecture for the micro-module computer room based on the thermal-electric synergistic optimization objective includes: Based on the aforementioned thermal-electric synergistic optimization objective, the monitoring requirements of the micro-module computer room are analyzed; Based on the aforementioned monitoring requirements, a sensor array for the micro-module computer room is constructed to integrate the multi-dimensional sensing layer of the micro-module computer room. Historical operational data of the micro-module data center is obtained to construct an operational status analysis model for the micro-module data center; Based on the aforementioned operational status analysis model, the operational status analysis layer of the micro-module computer room is integrated; Based on the aforementioned thermal-electric synergistic optimization objective, the objective function and optimization constraints of the micro-module computer room are defined. Based on the objective function and the optimization constraints, an optimization parameter calculation layer for the micro-module computer room is constructed. Define the PLC control algorithm for the micro-module computer room to construct the collaborative optimization execution layer of the micro-module computer room.

4. The micro-module computer room thermal-electric synergistic optimization method based on multi-parameter fusion control as described in claim 3, characterized in that, The objective function and optimization constraints for the micro-module computer room are defined based on the thermal-electric synergistic optimization objective, including: Based on the aforementioned thermal-electric synergistic optimization objective, the core sub-objectives of the micro-module data center are defined, wherein the core sub-objectives include: economic objective, PUE objective, cooling efficiency objective, and IT load balancing objective; Construct sub-objective functions for each of the core sub-objectives; The sub-objective functions are weighted and fused to obtain the objective function; Based on the equipment parameters of the micro-module computer room, the equipment safety parameters of the micro-module computer room are determined, wherein the equipment safety parameters include: air inlet temperature parameters, equipment power consumption parameters, power supply capacity parameters, and cooling capacity parameters; Define the equipment operation constraints of the micro-module computer room, wherein the equipment operation constraints include: air conditioning equipment constraints, load migration constraints, operation stability constraints, and operation performance constraints; Based on the equipment operation constraints and the equipment safety parameters, the optimized constraints for the micro-module computer room are constructed.

5. The micro-module computer room thermal-electric synergistic optimization method based on multi-parameter fusion control as described in claim 1, characterized in that, The process of fusing the multi-source heterogeneous data to obtain fused data includes: Identify the physical entities, logical entities, and conceptual entities of the micro-module computer room corresponding to the multi-source heterogeneous data; Define the entity relationships between the physical entity, the logical entity, and the conceptual entity; Based on the entity relationships, physical entities, logical entities, and conceptual entities, a knowledge graph framework for the micro-module computer room is constructed. The multi-source heterogeneous data is preprocessed to obtain preprocessed multi-source data; The preprocessed multi-source data is spatiotemporally aligned to obtain aligned multi-source data; The aligned multi-source data is encapsulated using event-based methods to obtain a multi-event object; Analyze the causal relationships between the multiple event objects and extract the multi-event features of the multiple event objects; Based on the causal relationship and the multi-event features, the multi-event objects are mapped to the knowledge graph framework to obtain fused data.

6. The micro-module computer room thermal-electrical synergistic optimization method based on multi-parameter fusion control as described in claim 1, characterized in that, The calculation of key status indicators of the micro-module data center based on the fused data and through the operational status analysis layer includes: The fused data is input into the operation status analysis model in the operation status analysis layer, and the operation status of the micro-module computer room is output. Extract the air intake temperature and power consumption of each server in the operating state to calculate the average temperature and temperature standard deviation of the micro-module computer room; Calculate the heat distribution coefficient of the micro-module computer room based on the average temperature and the temperature standard deviation; Extract the cold aisle temperature, hot aisle temperature, actual supply air temperature, and actual return air temperature from the operating state; Based on the cold aisle temperature, the hot aisle temperature, the actual supply air temperature, and the actual return air temperature, calculate the local airflow mixing degree and the global heat recovery efficiency of the micro-module computer room; Based on the local airflow mixing degree and the global heat recovery efficiency, the airflow organization effectiveness index of the micro-module computer room is calculated; Extract the total instantaneous power consumption of the cooling equipment and the total instantaneous heat dissipation power of the IT equipment in the operating state to calculate the total cooling load of the micro-module computer room; Based on the total cooling load, analyze the standard energy efficiency ratio of the micro-module computer room; Calculate the overall energy efficiency ratio of the cooling system of the micro-module computer room based on the total cooling load and the standard energy efficiency ratio; Extract the time series data of total IT power consumption from the fused data; Based on the total IT power consumption time series data, calculate the dynamic change rate of IT load in the micro-module computer room; Based on the heat distribution coefficient, the airflow organization effectiveness index, the overall energy efficiency ratio of the cooling system, and the dynamic change rate of the IT load, the key status indicators of the micro-module computer room are integrated.

7. The micro-module computer room thermal-electric synergistic optimization method based on multi-parameter fusion control as described in claim 6, characterized in that, The step of calculating the dynamic rate of change of IT load in the micro-module data center based on the total IT power consumption time series data includes: Based on the total IT power consumption time series data, calculate the logarithmic rate of return of the micro-module data center; Based on the total IT power consumption time series data, the fluctuation frequency and fluctuation pattern of the micro-module computer room are analyzed; Based on the fluctuation frequency and fluctuation pattern, the volatility of the micro-module data center is analyzed; The dynamic rate of change of IT load in the micro-module data center is calculated based on the volatility and the logarithmic return.

8. The micro-module computer room thermal-electric synergistic optimization method based on multi-parameter fusion control as described in claim 1, characterized in that, Based on the key state indicators, the objective function in the thermal-electric co-optimization network is solved through the optimization parameter calculation layer to obtain the co-optimization parameter sequence, including: Based on the key status indicators, the multi-timescale analysis data of the micro-module computer room corresponding to the key status indicators is analyzed through the optimization parameter calculation layer. Based on the multi-timescale analysis data and the objective function of the thermal-electric co-optimization network, a thermal-electric co-optimization model for the micro-module computer room is constructed. Based on the optimization constraints of the thermal-electric synergistic optimization network, the decision variable values ​​of the micro-module computer room are calculated using the thermal-electric synergistic optimization model. Based on the values ​​of the decision variables, a sequence of collaborative optimization parameters for the micro-module computer room is generated.

9. The micro-module computer room thermal-electrical synergistic optimization method based on multi-parameter fusion control as described in claim 8, characterized in that, The step of constructing the thermal-electric co-optimization model for the micro-module computer room based on the multi-timescale analysis data and the objective function of the thermal-electric co-optimization network includes: Based on the multi-timescale analysis data, the decision variables for the micro-module data center are determined, including: set temperature, fan speed, compressor frequency, power supply, and IT load migration. Construct a mapping relationship between the multi-timescale analysis data and the decision variables; Based on the mapping relationship and the objective function, an initial optimization model for the micro-module computer room is constructed; The initial optimization model is trained using a pre-set training set to obtain a trained optimization model; Analyze the accuracy of the initial training and optimization model; When the accuracy reaches a preset accuracy threshold, the trained and optimized model is used as the thermal-electric collaborative optimization model for the micro-module computer room.

10. The micro-module computer room thermal-electrical synergistic optimization method based on multi-parameter fusion control as described in claim 1, characterized in that, The step of generating collaborative optimization instructions for the micro-module data center based on the collaborative optimization parameter sequence through the collaborative optimization execution layer includes: Determine the instruction mapping rules for the collaborative optimization execution layer; The parameter sequence of the collaborative optimization is analyzed to obtain parameter analysis data; According to the instruction mapping rules, the parameter parsing data is converted into initial instructions through the PLC control algorithm of the collaborative optimization execution layer; Based on the timing requirements of the collaborative optimization parameter sequence, the execution time and execution order of the initial instructions are determined; Based on the execution time and the execution order, the initial instructions are formatted to obtain collaborative optimization instructions.

Images