Data center low-carbon control method and system

Abstract

The invention discloses a data center low-carbon control method and system, and relates to the technical field of data centers, and the method comprises the steps: collecting the environment parameters and equipment operation data of a data center in a hot-summer and warm-winter region in real time based on a multi-mode sensor network; constructing a climate-adaptive space-time diagram convolutional network, taking a server rack, a cooling tower and a power distribution cabinet as nodes, taking a cold and hot flow path as a dynamic weighted edge, embedding climate feature vectors, and generating a thermal load prediction map; a dynamic refrigeration strategy and a server load migration strategy are generated by taking a thermodynamic equation and a fluid dynamic model as strategy optimization boundary conditions through a physical constraint reinforcement learning framework and combining a thermal load prediction map; and based on the dynamic refrigeration strategy, the air conditioning system is controlled to start a hybrid refrigeration mode according to the real-time climate type, the water cooling main cycle and indirect evaporative cooling standby loop are synchronously operated, and the cold charging and discharging logic of the chilled water storage system is switched according to the electricity price time period.

Description

Technical Field

[0001] The present application relates to the field of data center technology, and in particular to a low-carbon control method and system for a data center. Background Art

[0002] In recent years, with the rapid development of the internet, services like big data and cloud computing have emerged, and the demand for large-scale data centers—the foundation for these developments—has also continued to grow. As the fundamental platform for carrying, transmitting, and computing power, data centers are the physical foundation for the application of new-generation digital technologies such as artificial intelligence, cloud computing, and blockchain. They are a crucial component of national strategic development and have become a key component of new infrastructure development.

[0003] With the rapid development of the digital economy, the number and scale of data centers are constantly expanding, resulting in significant energy consumption and carbon emissions. This is particularly true in regions with hot summers and warm winters, such as Guangzhou and Shenzhen, where the high temperature and humidity persist for extended periods, resulting in low utilization of natural cooling sources. This climate poses significant challenges to improving data center energy efficiency and controlling carbon emissions.

[0004] While research into key technologies for data center design and construction is actively underway both domestically and internationally, and some progress has been made, data center design and construction still face numerous challenges. For example, traditional data center cooling systems rely on fixed cooling modes, failing to fully account for dynamic climate changes and the real-time demands of equipment operation, resulting in low energy efficiency.

[0005] To address the above-mentioned problems, no effective solutions have been proposed so far. Summary of the Invention

[0006] The embodiments of the present application provide a data center low-carbon control method and system to solve the above technical problems.

[0007] This application provides a data center low-carbon control method, including:

[0008] Real-time data collection of environmental parameters and equipment operation data for data centers in hot summer and warm winter regions based on a multimodal sensor network. The environmental parameters include indoor and outdoor temperature and humidity gradients, heat flux density distribution, and the rate of change of the pressure difference between hot and cold aisles.

[0009] A climate-adaptive spatiotemporal graph convolutional network is constructed, using server racks, cooling towers, and power distribution cabinets as nodes, heat and cold flow paths as dynamically weighted edges, and climate feature vectors embedded to generate a heat load prediction map.

[0010] Through the physical constraint reinforcement learning framework, thermodynamic equations and fluid dynamics models are used as policy optimization boundary conditions, combined with the heat load prediction map to generate dynamic cooling strategies and server load migration strategies;

[0011] Based on the dynamic cooling strategy, the air-conditioning system is controlled to start the hybrid cooling mode according to the real-time climate type, synchronously operate the water cooling main loop and the indirect evaporative cooling backup loop, and switch the charging and discharging logic of the water storage system according to the electricity price period.

[0012] Furthermore, the climate-adaptive spatiotemporal graph convolutional network includes:

[0013] Climate feature embedding layer: Encodes seasonal humidity fluctuations in hot summer and warm winter regions and air pressure changes during typhoons into climate feature vectors and combines them with the real-time power data of device nodes.

[0014] Dynamic edge weight calculation module: dynamically adjusts edge weight coefficients based on the temperature rise rate, pressure difference change, and path length of the cold and hot flow paths, where the climate adaptation coefficient is generated through training of historical temperature and humidity data;

[0015] Thermodynamic attention mechanism: Introducing heat conduction equation constraints in the convolutional layer to suppress feature propagation that violates local energy conservation.

[0016] Furthermore, the physical constraint reinforcement learning framework includes:

[0017] The heat load prediction map, equipment energy efficiency ratio and pipeline pre-buried deviation data during the construction phase are integrated to define a state space;

[0018] The air conditioning cooling capacity adjustment range is limited to not exceed the critical condensation threshold calculated by the fluid dynamics model, and the server load migration path meets the redundant power supply safety rules;

[0019] A hierarchical reward function is designed, with short-term rewards based on the real-time PUE value and the temperature difference between hot and cold channels, and long-term rewards based on the total carbon emissions over the entire life cycle and the deviation from the construction progress.

[0020] Furthermore, the hybrid cooling mode prioritizes cooling high-density computing cluster areas through a dynamic refrigerant flow allocation algorithm, and utilizes waste heat from the liquid cooling circuit to drive local airflow circulation;

[0021] The control of the air conditioning system to start the mixed cooling mode according to the real-time climate type includes:

[0022] In high temperature and high humidity environments, the main water cooling circuit adopts a variable flow pump control strategy to adjust the branch valve opening according to the heat load distribution predicted by the spatiotemporal graph convolutional network;

[0023] When the indirect evaporative cooling backup circuit is started, weather forecast data is introduced to dynamically adjust the spraying frequency of the evaporative water curtain and the fresh air mixing ratio;

[0024] The charging and discharging logic of the water storage system is determined based on the electricity price time-sharing signal and the cooling tower efficiency curve. It prioritizes cold storage during periods of low electricity prices and triggers cooling when the cooling tower efficiency falls below a first set threshold.

[0025] Furthermore, the method further comprises:

[0026] Combined with the BIM twin model during the construction phase, reverse optimization of the refrigeration equipment layout and pipeline pre-buried was carried out to reduce cooling loss during the construction period;

[0027] The reverse optimization of the BIM twin model includes:

[0028] Identify refrigeration pipe installation errors through deviation analysis between laser scanning point cloud and design model;

[0029] Use topology optimization algorithms to re-plan pipeline routes to ensure the shortest refrigerant transportation distance and that the pressure drop complies with the fluid model constraints;

[0030] The optimized pipeline layout is reverse updated to the design model, and modular prefabricated component processing instructions are generated.

[0031] Furthermore, the realization of the liquid cooling circuit waste heat driving the airflow circulation includes:

[0032] A thermoelectric conversion module is integrated on the surface of the liquid cooling plate to convert waste heat into electricity to drive the micro-turbofan;

[0033] Adjust the turbofan speed based on infrared thermal imaging data to match the local airflow speed with the rack's cooling requirements;

[0034] When the ambient humidity exceeds the second set threshold, it automatically switches to anti-condensation mode, limits the maximum speed of the turbofan and starts auxiliary dehumidification.

[0035] Furthermore, the method further comprises:

[0036] Associating carbon measurement coding rules in the BIM twin model to record in real time the carbon emissions from concrete pouring during the construction phase and the amount of refrigerant leakage during the operation and maintenance phase;

[0037] Use the Hidden Markov Model to predict the evolution path of the carbon footprint over the entire life cycle, and dynamically optimize the cooling strategy and equipment replacement cycle;

[0038] When carbon emissions are predicted to exceed the threshold, equipment energy efficiency upgrades or renewable energy procurement plans are automatically triggered.

[0039] Furthermore, the method also includes an adaptive fault tolerance mechanism:

[0040] Pre-train equipment failure response strategies for high humidity scenarios during typhoons in a physical field model;

[0041] When the humidity sensor detects a sudden change in value, the humidity and heat compensation solution in the reserve strategy library is triggered, including increasing the cooling tower fan speed, migrating the edge computing load to the low humidity area rack, and starting the backup dehumidification unit.

[0042] Furthermore, the method further comprises:

[0043] Integrate micro heat sink fins and air flow guide grooves into lighting fixtures to guide LED waste heat to designated heat dissipation areas;

[0044] Dynamically adjust the inclination and brightness of lamps based on the rack heat distribution predicted by the spatiotemporal graph convolutional network to coordinate the heat dissipation airflow with the air supply path of the air conditioner;

[0045] During maintenance work, activate the work area lighting and heat dissipation enhancement mode to simultaneously increase local cooling capacity.

[0046] This application provides a data center low-carbon control system, including:

[0047] The environmental and equipment parameter acquisition module is used to collect environmental parameters and equipment operation data of data centers in hot summer and warm winter areas in real time based on a multimodal sensor network. The environmental parameters include indoor and outdoor temperature and humidity gradients, heat flux density distribution, and the rate of change of the pressure difference between hot and cold aisles.

[0048] The heat load prediction map generation module is used to build a climate-adaptive spatiotemporal graph convolutional network. It uses server racks, cooling towers, and power distribution cabinets as nodes, and heat and cold flow paths as dynamically weighted edges. It embeds climate feature vectors to generate heat load prediction maps.

[0049] A cooling and load migration strategy generation module is used to generate dynamic cooling strategies and server load migration strategies by combining the heat load prediction map with thermodynamic equations and fluid dynamics models as strategy optimization boundary conditions through a physical constraint reinforcement learning framework;

[0050] The cooling mode switching module is used to control the air-conditioning system to start the hybrid cooling mode according to the real-time climate type based on the dynamic cooling strategy, synchronously operate the water-cooling main loop and the indirect evaporative cooling backup loop, and switch the charging and discharging logic of the water storage system according to the electricity price period.

[0051] Based on the embodiments provided in this application, by deploying a multimodal sensor network, multi-dimensional parameters such as indoor and outdoor temperature and humidity gradients, heat flux density distribution, and the pressure difference change rate of hot and cold channels are collected in real time, breaking through the limitations of traditional fixed sensor layouts. The network can capture the dynamic changing characteristics of the microclimate inside the data center, provide high-resolution data support for subsequent strategy generation, and significantly improve the spatiotemporal accuracy of heat load prediction. The constructed spatiotemporal graph convolutional network uses key equipment as nodes and hot and cold flow paths as dynamically weighted edges, and embeds climate feature vectors into the graph structure. This dynamic modeling method can adjust the network weights in real time to adapt to the climate characteristics of large day and night temperature differences and frequent humidity fluctuations in hot summer and warm winter areas. Compared with traditional static prediction models, this method can predict the risk of local thermal runaway 2-3 hours in advance, reserving response time for cooling strategy optimization. By using thermodynamic equations and fluid dynamics models as boundary conditions for reinforcement learning, the present invention achieves a deep integration of physical laws and artificial intelligence. The framework can automatically generate dynamic cooling strategies and server load migration strategies, and dynamically optimize the PUE value while meeting the thermal safety of the equipment; based on real-time climate type judgment, the air-conditioning system can seamlessly switch between the water-cooled main loop and the indirect evaporative cooling backup loop. Especially during peak and valley periods of electricity prices, by optimizing the charging and discharging logic of the water storage cooling system, the peak of cooling energy consumption can be reduced and the operating costs can be reduced. This composite cooling mode demonstrates excellent energy efficiency in high temperature and high humidity environments. Through real-time heat load prediction and dynamic strategy generation, this method can significantly reduce the dependence of data centers on fossil energy. Under typical operating conditions in hot summer and warm winter areas, the average annual carbon emission intensity can be significantly reduced, providing a practical technical path for data centers to achieve carbon neutrality goals. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] The drawings described herein are used to provide a further understanding of the embodiments of the present invention and constitute a part of this application. The illustrative embodiments of this application and their descriptions are used to explain this application and do not constitute an improper limitation of this application. In the drawings:

[0053] Figure 1 is a flow chart of an optional low-carbon control method for a data center according to an embodiment of the present application;

[0054] Figure 2 is a flow chart of another optional low-carbon control method for a data center according to an embodiment of the present application;

[0055] Figure 3 This is a structural diagram of an optional data center low-carbon control system according to an embodiment of the present application.

[0056] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0057] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0058] At present, both domestic and foreign countries are actively carrying out research on key technologies for data center construction, such as intelligence, automation, energy saving, etc., and have achieved certain results. However, there is still a lot of room for improvement in the energy efficiency, site selection (different climatic conditions and different terrain conditions), space utilization, safety, reliability, and rapid construction of data centers. Therefore, this application mainly focuses on the impact of complex working environment conditions on the design, and the related research on reducing data PUE energy consumption indicators and rapid construction. With the rapid development of the digital economy, the demand for data storage and data center computing power in all walks of life continues to increase, and the scale of data centers continues to expand. Data centers mainly rely on server racks to accommodate IT equipment. The number of racks is an important indicator of the scale of data centers. High computing power requirements bring about problems of high energy consumption and high heat dissipation. The energy consumption of data center equipment and its supporting refrigeration and heat dissipation energy consumption have also risen.

[0059] The applicant's goal is to improve the efficiency and reduce carbon emissions during the design and construction of super-large data centers in hot summer and warm winter areas. By studying the key technologies for increasing efficiency and reducing carbon emissions in super-large data centers in hot summer and warm winter areas and their applications, the energy consumption of super-large data centers in the design and construction process in hot summer and warm winter areas can be reduced, space utilization, construction efficiency and construction quality can be improved, data center construction technology can be continuously improved, and the development of the national data center industry can be accelerated. By collecting different environmental working conditions in hot summer and warm winter areas, from the beginning of planning and design, it is necessary to fully consider related issues such as project energy conservation and consumption reduction, and reduction of operation and maintenance costs, so that the overall PUE value reaches the leading level of similar projects in the same latitude areas around the world, and at the same time improve the automation operation and maintenance level of the low-carbon control system of the data center and reduce operation and maintenance costs. Specific research contents include:

[0060] The research on low-carbon design of ultra-large data centers in complex environmental conditions mainly includes the following aspects:

[0061] (1) Research on the design of air conditioning and refrigeration technology for ultra-large data centers under different climatic conditions.

[0062] The northern climate is characterized by dryness and water scarcity, low annual average temperatures, and limited water resources. Therefore, data center projects typically use air cooling for air conditioning and cooling, leveraging the lower temperatures while mitigating the drawbacks of water scarcity. The southern climate, characterized by high temperatures and higher average temperatures than the north, is characterized by abundant water resources. Therefore, data center projects typically use water cooling for air conditioning and cooling. Air cooling requires a semi-open, unenclosed building envelope, with the air cooling source acting as the medium to enter the air conditioning unit. Water cooling, on the other hand, does not require an open building envelope and places no special requirements on the envelope. Research is needed to determine the climate characteristics and average temperatures that are most effective for air cooling, the water resource environment characteristics that are most suitable for water cooling, and the impact of different air conditioning and cooling technologies on the overall functional design of the building.

[0063] (2) Research on the functional layout planning and design of ultra-large data centers under different terrain conditions.

[0064] The functional layout of the data center's plan location affects the functional layout of the building, and the cost of outdoor roads and outdoor pipeline projects.

[0065] (3) Research on key design technologies for reducing PUE in ultra-large data centers under complex environmental conditions.

[0066] The PUE (Power Usage Effectiveness) value is an internationally accepted metric for measuring data center power usage efficiency. The closer the PUE value is to 1, the more green a data center is. According to relevant Chinese regulations, data centers in the Shaoguan cluster must reduce their PUE value to below 1.25, striving to create a green and low-carbon data center cluster. By establishing an energy consumption calculation model for ultra-large data centers in hot summer and warm winter regions, energy consumption indicators are analyzed and the factors affecting energy consumption indicators are summarized. Based on the factors affecting energy consumption indicators, different energy-saving technical measures are implemented and compared and analyzed to identify the optimal carbon-reducing and energy-saving technical measures (equipment energy saving, water cooling, and energy storage). During the operational phase, computing power is a dynamic process, and the number of racks is not deployed at peak all at once. Therefore, the PUE variation patterns during the dynamic process of different rack numbers (power) are worthy of discussion and research.

[0067] Research on carbon reduction applications of BIM technology in ultra-large data center projects, including:

[0068] Against the backdrop of global climate change, reducing carbon emissions has become a critical task facing all industries. As core infrastructure of the information age, ultra-large data centers pose a particularly prominent carbon emission challenge during their construction and operation. Therefore, research on the application of BIM technology in ultra-large data center projects to reduce carbon emissions is crucial for promoting the green and sustainable development of the data center industry.

[0069] This study aims to explore how to effectively reduce carbon emissions throughout the entire lifecycle of ultra-large data center projects, including design, detailed design, and construction, through the application of BIM technology. Specifically, we will focus on the application of BIM technology in optimizing design during the design phase and implementing refined management during the construction phase, with the goal of achieving carbon reduction goals.

[0070] (1) Research on the application of BIM technology in carbon reduction during the design phase of ultra-large data center projects.

[0071] During the design phase, BIM technology's precise modeling and simulation analysis capabilities are used to study and evaluate the energy consumption and carbon emissions of ultra-large data center projects, enabling optimization from the early stages of design. BIM technology is used to optimize design parameters such as building orientation, window-to-wall ratio, and building form to reduce solar radiation and heat transfer, thereby lowering building energy consumption and carbon emissions. Furthermore, BIM technology is used to dynamically simulate and predict building energy consumption for ultra-large data center projects, providing real-time energy consumption data feedback and enabling timely adjustments to design solutions to ensure that projects meet functional requirements while achieving the lowest energy consumption and carbon emissions.

[0072] (2) Research on carbon reduction applications of BIM technology in the in-depth design of ultra-large data center projects.

[0073] During the in-depth design phase of ultra-large data center projects, we study and apply BIM technology for fine modeling, conduct comprehensive pipeline layout and collision detection, realize equipment layout and management, professional coordination, pipeline integration, clearance control, parameter review, support and hanger design, and electromechanical terminal and reserved embedded positioning, and output electromechanical pipeline comprehensive drawings, electromechanical professional construction in-depth design drawings, and related professional coordination condition drawings, effectively optimizing equipment layout, pipeline direction and material use, reducing design changes and material waste, and thus reducing carbon emissions.

[0074] (3) Research on the carbon reduction application of BIM technology in the construction phase of ultra-large data center projects.

[0075] By establishing a BIM model and applying BIM technology to carbon emissions statistics during the construction phase of a hyperscale data center project, we can accurately track and record carbon emissions data during the construction process, including energy consumption of construction equipment and carbon emissions from the transportation and processing of materials. The BIM model provides powerful data integration and analysis capabilities, making the management and reporting of carbon emissions data more accurate and efficient. Through real-time monitoring and analysis of carbon emissions data, we can promptly identify carbon emission hotspots and potential improvement points, and implement appropriate measures to reduce emissions.

[0076] Alternatively, as Figure 1 As shown, the present application provides a data center low-carbon control method, comprising:

[0077] S101 collects environmental parameters and equipment operation data of data centers in hot summer and warm winter areas in real time based on a multimodal sensor network. Environmental parameters include indoor and outdoor temperature and humidity gradients, heat flux density distribution, and the rate of change of the pressure difference between hot and cold aisles.

[0078] S102: Build a climate-adaptive spatiotemporal graph convolutional network with server racks, cooling towers, and power distribution cabinets as nodes, heat and cold flow paths as dynamically weighted edges, embed climate feature vectors, and generate a heat load prediction map.

[0079] S103 uses a physical constraint reinforcement learning framework to use thermodynamic equations and fluid dynamics models as policy optimization boundary conditions, combined with heat load prediction maps, to generate dynamic cooling strategies and server load migration strategies.

[0080] S104, based on the dynamic cooling strategy, controls the air conditioning system to start the hybrid cooling mode according to the real-time climate type, synchronously operates the water cooling main loop and the indirect evaporative cooling backup loop, and switches the charging and discharging logic of the water storage cooling system according to the electricity price period.

[0081] Based on the embodiments provided in this application, by deploying a multimodal sensor network, multi-dimensional parameters such as indoor and outdoor temperature and humidity gradients, heat flux density distribution, and the pressure difference change rate of hot and cold channels are collected in real time, breaking through the limitations of traditional fixed sensor layouts. The network can capture the dynamic changing characteristics of the microclimate inside the data center, provide high-resolution data support for subsequent strategy generation, and significantly improve the spatiotemporal accuracy of heat load prediction. The constructed spatiotemporal graph convolutional network uses key equipment as nodes and hot and cold flow paths as dynamically weighted edges, and embeds climate feature vectors into the graph structure. This dynamic modeling method can adjust the network weights in real time to adapt to the climate characteristics of large day and night temperature differences and frequent humidity fluctuations in hot summer and warm winter areas. Compared with traditional static prediction models, this method can predict the risk of local thermal runaway 2-3 hours in advance, reserving response time for cooling strategy optimization. By using thermodynamic equations and fluid dynamics models as boundary conditions for reinforcement learning, the present invention achieves a deep integration of physical laws and artificial intelligence. The framework can automatically generate dynamic cooling strategies and server load migration strategies, and dynamically optimize the PUE value while meeting the thermal safety of the equipment; based on real-time climate type judgment, the air-conditioning system can seamlessly switch between the water-cooled main loop and the indirect evaporative cooling backup loop. Especially during peak and valley periods of electricity prices, by optimizing the charging and discharging logic of the water storage cooling system, the peak of cooling energy consumption can be reduced and the operating costs can be reduced. This composite cooling mode demonstrates excellent energy efficiency in high temperature and high humidity environments. Through real-time heat load prediction and dynamic strategy generation, this method can significantly reduce the dependence of data centers on fossil energy. Under typical operating conditions in hot summer and warm winter areas, the average annual carbon emission intensity can be significantly reduced, providing a practical technical path for data centers to achieve carbon neutrality goals.

[0082] Furthermore, the climate-adaptive spatiotemporal graph convolutional network includes:

[0083] Climate feature embedding layer: Encodes seasonal humidity fluctuations in hot summer and warm winter regions and air pressure changes during typhoons into climate feature vectors and combines them with the real-time power data of device nodes.

[0084] Dynamic edge weight calculation module: Dynamically adjusts edge weight coefficients based on the temperature rise rate, pressure difference change, and path length of the cold and hot flow paths. The climate adaptation coefficient is generated through training of historical temperature and humidity data.

[0085] Thermodynamic attention mechanism: Introducing heat conduction equation constraints in the convolutional layer to suppress feature propagation that violates local energy conservation.

[0086] In the embodiment of the present application, the climate-adaptive dynamic edge weight is calculated based on the following formula:

[0087]

[0088] Among them, W {ij} is the edge weight from node i to j, reflecting the priority of the cold and hot flow paths; α(T {season} ) is the seasonal humidity adaptive coefficient, which is generated based on historical temperature and humidity data and is used to amplify / suppress the impact of temperature rise; ΔT {ij} is the temperature rise rate from node i to j (℃ / s), and the temperature difference between hot and cold streams is monitored in real time. typhoon ) is the pressure adaptation coefficient during the typhoon period, encoding the pressure fluctuation characteristics under high humidity during the typhoon period; ΔP {ij} is the pressure difference change between the cold and hot flow paths (Pa), reflecting the dynamics of airflow resistance; L {ij} is the path length (m), which optimizes the refrigerant delivery efficiency; Δt is the time interval. In the data center scenario, Δt can be expressed as the time step for the sensor to collect data, such as collecting data every minute or every 5 minutes. The value of Δt depends on the system's real-time requirements and is usually in the range of 1-10 minutes; α is a coefficient related to seasonal humidity, which is used to adjust the impact of the temperature rise rate on the edge weight. Its value range is usually between 0.5 and 1.5. The specific value can be generated through training of historical temperature and humidity data. For example: in seasons with low humidity (such as winter), α can be taken as 0.8 to suppress the impact of temperature rise; in seasons with high humidity (such as summer), α can be taken as 1.2 to amplify the impact of temperature rise; β is a coefficient related to air pressure during typhoon period, which is used to adjust the impact of pressure difference changes on edge weights. Its value range is usually between 0.5 and 1.5. The specific value can be generated through training of historical air pressure data. For example, during typhoon period, when the air pressure fluctuations are large, β can be set to 1.5 to amplify the impact of pressure difference changes; during non-typhoon period, when the air pressure fluctuations are small, β can be set to 0.8 to suppress the impact of pressure difference changes.

[0089] Based on the embodiments provided in this application, the climate feature embedding layer and the dynamic edge weight calculation module, by encoding seasonal humidity fluctuations and typhoon period air pressure changes, enable the climate-adaptive spatiotemporal graph convolutional network to accurately predict the heat load distribution under sudden climate changes, reducing the heat load prediction error compared to traditional static graph models.

[0090] Furthermore, the physical constraint reinforcement learning framework includes:

[0091] Integrate heat load prediction maps, equipment energy efficiency ratios, and pipeline pre-buried deviation data during the construction phase to define the state space;

[0092] Limit the air conditioning cooling capacity adjustment range to no more than the critical condensation threshold calculated by the fluid dynamics model, and ensure that the server load migration path meets the redundant power supply safety rules;

[0093] A hierarchical reward function is designed, with short-term rewards based on the real-time PUE value and the temperature difference between hot and cold channels, and long-term rewards based on the total carbon emissions over the entire life cycle and the deviation from the construction progress.

[0094] In this embodiment of the present application, the physical constraint reinforcement learning hierarchical reward function is:

[0095] R=γ1×(1-PUE t )+γ2×CDR t -γ3×Delay construction

[0096] Among them, R is the comprehensive reward value, which is used for multi-objective balance of strategy optimization; PUE t Real-time power efficiency is calculated through sensor networks; CDR t Real-time carbon emission intensity (kgCO2 / kWh), related to refrigerant leakage and equipment energy consumption; Delay construction is the penalty term for construction progress deviation, which is dynamically calculated based on the BIM model process conflict detection results; γ1, γ2, and γ3 are weight coefficients, which are adaptively adjusted based on historical operation and maintenance data;

[0097] For example, in a high-energy-consuming data center scenario, the goal is to reduce energy consumption, reduce carbon emissions, and ensure construction progress. The weight coefficient values ​​are as follows: γ1 = 0.5: Real-time power usage effectiveness (PUE) has a greater impact on energy consumption, so it is given a higher weight. γ2 = 0.3: Carbon emission intensity (CDR) has a significant impact on the environment, but in this scenario it is secondary to energy consumption optimization. γ3 = 0.2: Construction schedule deviation (Delay_construction) has a smaller impact on the overall project and has a lower weight. In a high-energy-consuming data center, reducing energy consumption is the top priority, so PUE has a higher weight. At the same time, carbon emission intensity also needs attention, but it is relatively secondary. The impact of construction schedule deviation is relatively small, so it has the lowest weight.

[0098] In key carbon emission regulatory areas, the goal is to prioritize reducing carbon emissions while also taking into account energy consumption and construction progress. Weight coefficient values ​​are: γ1 = 0.3: The impact of real-time power usage effectiveness (PUE) on energy consumption is moderately weighted. γ2 = 0.5: Carbon emission intensity (CDR) is the primary optimization objective and has the highest weight. γ3 = 0.2: Construction schedule deviation (Delay_construction) has a lower weight. In key carbon emission regulatory areas, reducing carbon emissions is the top priority, so CDR has the highest weight. PUE and construction schedule deviation have relatively lower weights, but still require attention.

[0099] In a data center with a tight construction schedule, the goal is to ensure on-time completion while balancing energy consumption and carbon emissions. Weighting coefficients are: γ1 = 0.3: Power Usage Effectiveness (PUE) is moderately weighted. γ2 = 0.3: Carbon Dioxide Reduction (CDR) is moderately weighted. γ3 = 0.4: Construction Schedule Deviation is weighted the highest. In a data center with a tight construction schedule, ensuring on-time project completion is the top priority, so Delay_construction is given the highest weight. PUE and CDR are weighted relatively lower, but still require attention.

[0100] Based on the embodiments provided in this application, a hierarchical reward function integrates real-time PUE, carbon emission intensity, and construction progress deviation to ensure that the strategy strikes a balance between short-term energy efficiency and long-term carbon footprint, addressing the limitations of single-objective optimization in existing technologies.

[0101] Furthermore, the hybrid cooling mode prioritizes cooling high-density computing cluster areas through a dynamic refrigerant flow allocation algorithm, and utilizes waste heat from the liquid cooling circuit to drive local airflow circulation;

[0102] like Figure 2 As shown, the air conditioning system is controlled to start the mixed cooling mode according to the real-time climate type, including:

[0103] S201, in a high temperature and high humidity environment, the main water cooling circuit adopts a variable flow pump control strategy to adjust the branch valve opening according to the heat load distribution predicted by the spatiotemporal graph convolutional network;

[0104] S202, when the indirect evaporative cooling backup circuit is started, weather forecast data is introduced to dynamically adjust the spraying frequency and fresh air mixing ratio of the evaporative water curtain;

[0105] S203, the charging and discharging logic of the water cooling system is determined based on the electricity price time-sharing signal and the cooling tower efficiency curve, with priority given to cooling during the low electricity price period, and cooling is triggered when the cooling tower efficiency falls below a first set threshold.

[0106] The first set threshold may include 0.7, 0.6, etc.

[0107] Based on the embodiments provided in this application, the water-cooling branch valve opening adjustment model introduces a humidity correction factor and meteorological forecast data to dynamically optimize the fresh air mixing ratio of the indirect evaporative cooling circuit and reduce the energy consumption of the cooling tower in a high temperature and high humidity environment.

[0108] Furthermore, the method further comprises:

[0109] Combined with the BIM twin model during the construction phase, reverse optimization of the refrigeration equipment layout and pipeline pre-buried was carried out to reduce cooling loss during the construction period;

[0110] Among them, the reverse optimization of the BIM twin model includes:

[0111] Identify refrigeration pipe installation errors through deviation analysis between laser scanning point cloud and design model;

[0112] Use topology optimization algorithms to re-plan pipeline routes to ensure the shortest refrigerant transportation distance and that the pressure drop complies with the fluid model constraints;

[0113] The optimized pipeline layout is reverse updated to the design model, and modular prefabricated component processing instructions are generated.

[0114] Based on the embodiments provided in this application, pipeline reverse optimization based on laser scanning is combined with a topological algorithm to re-plan the refrigerant path, reduce cooling loss during the construction period, and improve pipeline installation accuracy to the millimeter level.

[0115] The generating of modular prefabricated component processing instructions includes:

[0116] Decompose the piping system into prefabricated sections with standardized interfaces and pre-install temperature and pressure sensors in the prefabricated sections;

[0117] During construction, drone inspections are used to match prefabricated sections with on-site positioning coordinates, and the BIM twin model is combined to verify assembly accuracy in real time, achieving rapid assembly with an error less than a set threshold.

[0118] Furthermore, the realization of the airflow circulation driven by the waste heat of the liquid cooling circuit includes:

[0119] A thermoelectric conversion module is integrated on the surface of the liquid cooling plate to convert waste heat into electricity to drive the micro-turbofan;

[0120] Adjust the turbofan speed based on infrared thermal imaging data to match the local airflow speed with the rack's cooling requirements;

[0121] When the ambient humidity exceeds the second set threshold, it automatically switches to anti-condensation mode, limits the maximum speed of the turbofan and starts auxiliary dehumidification.

[0122] The second threshold can be set to a humidity value close to the condensation point, such as 90% RH. If a device has specific humidity requirements, such as certain servers requiring even lower humidity to avoid condensation, the threshold can be set between 85% and 90% RH to provide a safety margin. The threshold can be dynamically adjusted based on the actual operating environment and device cooling requirements. For example, in high-humidity environments, the threshold can be appropriately lowered to enable early entry into anti-condensation mode, ensuring device safety.

[0123] Based on the embodiments provided in the present application, a micro-turbofan system driven by waste heat from a liquid cooling circuit converts waste heat into airflow power through thermoelectric conversion, and automatically switches to an anti-condensation mode when the humidity exceeds a threshold, thereby avoiding equipment failures caused by sudden changes in humidity in traditional cooling solutions.

[0124] Furthermore, the method further comprises:

[0125] Associating carbon measurement coding rules in the BIM twin model to record in real time the carbon emissions from concrete pouring during the construction phase and the amount of refrigerant leakage during the operation and maintenance phase;

[0126] Use the Hidden Markov Model to predict the evolution path of the carbon footprint over the entire life cycle, and dynamically optimize the cooling strategy and equipment replacement cycle;

[0127] When carbon emissions are predicted to exceed the threshold, equipment energy efficiency upgrades or renewable energy procurement plans are automatically triggered.

[0128] In the embodiment of the present application, the hidden Markov carbon footprint state transition is calculated based on the following formula:

[0129] S {t+1} =S t ×A s +E construction ×B operate

[0130] Among them, S t is the carbon footprint state vector at time t, including the carbon emission components of the construction and operation and maintenance stages; A sis the state transfer matrix of the construction phase, which correlates the carbon emissions of processes such as concrete pouring and pipeline welding; E construction is the construction error correction factor, calculated based on the deviation of the laser scanning point cloud; B operate The impact matrix of the operation and maintenance phase maps the coupling effect of cooling strategy and equipment aging on carbon emissions; S {t+1} is the carbon footprint state vector at time t+1;

[0131] Based on the embodiments provided in this application, the hidden Markov carbon footprint model achieves dynamic tracking and prediction of carbon emissions throughout the life cycle by associating construction errors with operation and maintenance strategies, providing accurate decision-making support for carbon quota management.

[0132] Furthermore, the method also includes an adaptive fault tolerance mechanism:

[0133] Pre-train equipment failure response strategies for high humidity scenarios during typhoons in a physical field model;

[0134] When the humidity sensor detects a sudden change in value, the humidity and heat compensation solution in the reserve strategy library is triggered, including increasing the cooling tower fan speed, migrating the edge computing load to the low humidity area rack, and starting the backup dehumidification unit.

[0135] Based on the embodiments provided in this application, the typhoon period humidity and heat compensation mechanism pre-trains a variety of fault scenario response strategies, quickly migrates computing loads and starts backup dehumidification when humidity suddenly changes, ensuring continuous and stable operation of the system in extreme climates.

[0136] Furthermore, the method further comprises:

[0137] Integrate micro heat sink fins and air flow guide grooves into lighting fixtures to guide LED waste heat to designated heat dissipation areas;

[0138] Based on the rack heat distribution predicted by the spatiotemporal graph convolutional network, the lighting inclination and brightness are dynamically adjusted to ensure that the cooling airflow and the air supply path of the air conditioner are coordinated.

[0139] During maintenance work, activate the work area lighting and heat dissipation enhancement mode to simultaneously increase local cooling capacity.

[0140] Based on the embodiments provided in this application, lighting-heat dissipation coupling control converts LED waste heat into auxiliary heat dissipation resources through lamp structure optimization and thermal distribution linkage adjustment, thereby improving local cooling efficiency while reducing lighting energy consumption.

[0141] Alternatively, as Figure 3 As shown, the present application provides a data center low-carbon control system, comprising:

[0142] Environmental and equipment parameter acquisition module 301 is used to collect environmental parameters and equipment operation data of data centers in hot summer and warm winter areas in real time based on a multimodal sensor network. Environmental parameters include indoor and outdoor temperature and humidity gradients, heat flux density distribution, and the rate of change of the pressure difference between hot and cold aisles.

[0143] Heat load prediction map generation module 302 is used to build a climate-adaptive spatiotemporal graph convolutional network, using server racks, cooling towers, and power distribution cabinets as nodes, cold and hot flow paths as dynamically weighted edges, and embedding climate feature vectors to generate a heat load prediction map;

[0144] The cooling and load migration strategy generation module 303 is used to generate dynamic cooling strategies and server load migration strategies by using a physical constraint reinforcement learning framework, taking thermodynamic equations and fluid dynamics models as strategy optimization boundary conditions, and combining them with the heat load prediction map;

[0145] The cooling mode switching module 304 is used to control the air-conditioning system to start the hybrid cooling mode according to the real-time climate type based on the dynamic cooling strategy, synchronously operate the water cooling main loop and the indirect evaporative cooling backup loop, and switch the charging and discharging logic of the water storage cooling system according to the electricity price period.

[0146] It should be noted that in this application, the embodiments implemented on the data center low-carbon control system side can be referenced with the embodiments implemented on the data center low-carbon control method side, and this application will not describe them one by one.

[0147] The above are only preferred embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A data center low-carbon control method, characterized in that: include: Real-time data collection of environmental parameters and equipment operation data for data centers in hot summer and warm winter regions based on a multimodal sensor network. The environmental parameters include indoor and outdoor temperature and humidity gradients, heat flux density distribution, and the rate of change of the pressure difference between hot and cold aisles. A climate-adaptive spatiotemporal graph convolutional network is constructed, using server racks, cooling towers, and power distribution cabinets as nodes, heat and cold flow paths as dynamically weighted edges, and climate feature vectors embedded to generate a heat load prediction map. Through the physical constraint reinforcement learning framework, thermodynamic equations and fluid dynamics models are used as policy optimization boundary conditions, combined with the heat load prediction map to generate dynamic cooling strategies and server load migration strategies; Based on the dynamic cooling strategy, the air-conditioning system is controlled to start the hybrid cooling mode according to the real-time climate type, synchronously operate the water cooling main loop and the indirect evaporative cooling backup loop, and switch the charging and discharging logic of the water storage system according to the electricity price period.

2. The data center low-carbon control method according to claim 1, characterized in that: The climate-adaptive spatiotemporal graph convolutional network includes: Climate feature embedding layer: Encodes seasonal humidity fluctuations in hot summer and warm winter regions and air pressure changes during typhoons into climate feature vectors and combines them with the real-time power data of device nodes. Dynamic edge weight calculation module: dynamically adjusts edge weight coefficients based on the temperature rise rate, pressure difference change, and path length of the cold and hot flow paths, where the climate adaptation coefficient is generated through training of historical temperature and humidity data; Thermodynamic attention mechanism: Introducing heat conduction equation constraints in the convolutional layer to suppress feature propagation that violates local energy conservation.

3. The data center low-carbon control method according to claim 2, characterized in that: The physical constraint reinforcement learning framework includes: The heat load prediction map, equipment energy efficiency ratio and pipeline pre-buried deviation data during the construction phase are integrated to define a state space; The air conditioning cooling capacity adjustment range is limited to not exceed the critical condensation threshold calculated by the fluid dynamics model, and the server load migration path meets the redundant power supply safety rules; A hierarchical reward function is designed, with short-term rewards based on the real-time PUE value and the temperature difference between hot and cold channels, and long-term rewards based on the total carbon emissions over the entire life cycle and the deviation from the construction progress.

4. The data center low-carbon control method according to claim 3, characterized in that: The hybrid cooling mode prioritizes cooling high-density computing cluster areas through a dynamic refrigerant flow allocation algorithm and utilizes waste heat from the liquid cooling circuit to drive local airflow circulation. The control of the air conditioning system to start the mixed cooling mode according to the real-time climate type includes: In high temperature and high humidity environments, the main water cooling circuit adopts a variable flow pump control strategy to adjust the branch valve opening according to the heat load distribution predicted by the spatiotemporal graph convolutional network; When the indirect evaporative cooling backup circuit is started, weather forecast data is introduced to dynamically adjust the spraying frequency of the evaporative water curtain and the fresh air mixing ratio; The charging and discharging logic of the water storage system is determined based on the electricity price time-sharing signal and the cooling tower efficiency curve. It prioritizes cold storage during periods of low electricity prices and triggers cooling when the cooling tower efficiency falls below a first set threshold.

5. The data center low-carbon control method according to claim 1, characterized in that: The method further comprises: Combined with the BIM twin model during the construction phase, reverse optimization of the refrigeration equipment layout and pipeline pre-buried was carried out to reduce cooling loss during the construction period; The reverse optimization of the BIM twin model includes: Identify refrigeration pipe installation errors through deviation analysis between laser scanning point cloud and design model; Use topology optimization algorithms to re-plan pipeline routes to ensure the shortest refrigerant transportation distance and that the pressure drop complies with the fluid model constraints; The optimized pipeline layout is reverse updated to the design model, and modular prefabricated component processing instructions are generated.

6. The data center low-carbon control method according to claim 4, characterized in that: The realization of the liquid cooling circuit waste heat driven airflow circulation includes: A thermoelectric conversion module is integrated on the surface of the liquid cooling plate to convert waste heat into electricity to drive the micro-turbofan; Adjust the turbofan speed based on infrared thermal imaging data to match the local airflow speed with the rack's cooling requirements; When the ambient humidity exceeds the second set threshold, it automatically switches to anti-condensation mode, limits the maximum speed of the turbofan and starts auxiliary dehumidification.

7. The data center low-carbon control method according to claim 5, characterized in that: The method further comprises: Associating carbon measurement coding rules in the BIM twin model to record in real time the carbon emissions from concrete pouring during the construction phase and the amount of refrigerant leakage during the operation and maintenance phase; Use the Hidden Markov Model to predict the evolution path of the carbon footprint over the entire life cycle, and dynamically optimize the cooling strategy and equipment replacement cycle; When carbon emissions are predicted to exceed the threshold, equipment energy efficiency upgrades or renewable energy procurement plans are automatically triggered.

8. The data center low-carbon control method according to claim 1, characterized in that: The method also includes an adaptive fault tolerance mechanism: Pre-train equipment failure response strategies for high humidity scenarios during typhoons in a physical field model; When the humidity sensor detects a sudden change in value, the humidity and heat compensation solution in the reserve strategy library is triggered, including increasing the cooling tower fan speed, migrating the edge computing load to the low humidity area rack, and starting the backup dehumidification unit.

9. The data center low-carbon control method according to claim 1, characterized in that: The method further comprises: Integrate micro heat sink fins and air flow guide grooves into lighting fixtures to guide LED waste heat to designated heat dissipation areas; Dynamically adjust the inclination and brightness of lamps based on the rack heat distribution predicted by the spatiotemporal graph convolutional network to coordinate the heat dissipation airflow with the air supply path of the air conditioner; During maintenance work, activate the work area lighting and heat dissipation enhancement mode to simultaneously increase local cooling capacity.

10. A low-carbon control system for a data center, characterized in that: include: The environmental and equipment parameter acquisition module is used to collect environmental parameters and equipment operation data of data centers in hot summer and warm winter areas in real time based on a multimodal sensor network. The environmental parameters include indoor and outdoor temperature and humidity gradients, heat flux density distribution, and the rate of change of the pressure difference between hot and cold aisles. The heat load prediction map generation module is used to build a climate-adaptive spatiotemporal graph convolutional network. It uses server racks, cooling towers, and power distribution cabinets as nodes, and heat and cold flow paths as dynamically weighted edges. It embeds climate feature vectors to generate heat load prediction maps. A cooling and load migration strategy generation module is used to generate dynamic cooling strategies and server load migration strategies by combining the heat load prediction map with thermodynamic equations and fluid dynamics models as strategy optimization boundary conditions through a physical constraint reinforcement learning framework; The cooling mode switching module is used to control the air-conditioning system to start the hybrid cooling mode according to the real-time climate type based on the dynamic cooling strategy, synchronously operate the water-cooling main loop and the indirect evaporative cooling backup loop, and switch the charging and discharging logic of the water storage system according to the electricity price period.

Images