Machine room control method and modular data machine room

Abstract

The application relates to a machine room control method and a modular data machine room. The control method comprises the following steps: acquiring configuration information of a refrigeration equipment module and performing analysis to generate basic refrigeration capacity characteristic information of an equipment container; acquiring cabinet deployment information of a cabinet, processing the cabinet deployment information, and generating refrigeration demand characteristic information of the cabinet; performing matching calculation on the basic refrigeration capacity characteristic information and the refrigeration demand characteristic information to generate an initial refrigeration control strategy; acquiring environment state information collected by a monitoring and fire-fighting equipment module pre-integrated in the equipment container, correcting the initial refrigeration control strategy according to the environment state information to generate a dynamic corrected refrigeration control strategy; and controlling the operation state of the refrigeration equipment module based on the dynamic corrected refrigeration control strategy to control the refrigeration output of the prefabricated modular data machine room.

Description

Technical Field

[0001] This application relates to the technical field of IT data centers, and more particularly to a data center control method and a modular data center. Background Technology

[0002] With the continuous expansion of data center scale and the sustained increase in server power density, energy management of data center server rooms, especially the energy consumption control of cooling systems, has become a key issue of concern in the industry. To achieve green and energy-saving goals, it is necessary to precisely control the cooling capacity of data center server rooms to dynamically match it with the actual heat load generated by IT equipment, avoiding equipment risks and energy waste caused by insufficient or excessive cooling.

[0003] Currently, a common data center cooling control scheme relies primarily on the overall average temperature of the data center space as a feedback signal. Specifically, this scheme installs several temperature sensors within the data center to monitor the ambient temperature and compares the average temperature reading with a preset temperature threshold. When the average temperature exceeds the threshold, all cooling equipment is activated or its power is increased according to fixed logic; conversely, its power is reduced or some equipment is shut down. The entire process depends mainly on the response to a single ambient temperature parameter.

[0004] However, in the existing solutions mentioned above, the cooling capacity adjustment only responds to changes in the overall ambient temperature and cannot perceive or adapt to the differentiated and refined cooling needs generated by the number, distribution density, actual power consumption, and local hotspots within the data center. This results in a mismatch between the cooling capacity supply and the actual heat load generated by the server rack equipment in terms of space and magnitude, easily leading to insufficient cooling in some areas of the data center while excessive cooling in others, resulting in low overall cooling efficiency and energy utilization efficiency. Summary of the Invention

[0005] This application provides a data center control method and a modular data center, aiming to improve the overall cooling efficiency and energy utilization efficiency of the data center.

[0006] To address the aforementioned technical problems, this application provides the following technical solutions in its embodiments: A computer room control method, comprising: The configuration information of the refrigeration equipment modules pre-integrated within the equipment container is obtained, and the configuration information is parsed to generate basic refrigeration capacity characterization information of the equipment container. Obtain rack deployment information of one or more racks deployed on-site in the modular data center, process the rack deployment information to generate cooling demand characterization information for the one or more racks; The basic cooling capacity characterization information and the cooling demand characterization information are matched and calculated to generate an initial cooling control strategy; The system acquires environmental status information collected by the monitoring and fire protection equipment module pre-integrated inside the equipment container, and modifies the initial cooling control strategy based on the environmental status information to generate a dynamically modified cooling control strategy. Based on the aforementioned dynamic correction cooling control strategy, the operating status of the cooling equipment module is controlled to control the cooling output of the prefabricated modular data center.

[0007] On the other hand, embodiments of this application also provide the following technical solutions: A modular data center, capable of applying the aforementioned data center control methods, includes: Supporting devices for equipment containers, one or more cabinets, and low-voltage cable trays; The equipment container is pre-integrated with refrigeration equipment modules, power distribution equipment modules, and monitoring and fire protection equipment modules. The one or more cabinets are arranged independently of the equipment container and are detachably connected to the refrigeration equipment module and power distribution equipment module inside the equipment container; The support device is rotatably mounted on the top or side of the equipment container; The support device has a retracted state and an extended state: in the retracted state, the support device is rotated and stored in the internal space of the equipment container or close to the outer contour; in the extended state, the support device is rotated out from the equipment container and located above the one or more cabinets, for supporting and fixing the low-voltage cable tray.

[0008] On the other hand, a computer device is provided, the device including a processor and a memory, the memory storing at least one instruction or at least one program, the at least one instruction or the at least one program being loaded and executed by the processor to implement the computer room control method as described above.

[0009] On the other hand, a computer-readable storage medium is provided, wherein at least one instruction or at least one program is stored therein, the at least one instruction or the at least one program being loaded and executed by a processor to implement the computer room control method described above.

[0010] The beneficial effects of this application are as follows: 1. This application expands the control basis from a single ambient temperature to the cabinet information itself, representing the heat load source, by performing the step of "obtaining cabinet deployment information of one or more cabinets deployed on-site in a modular data center, and processing the cabinet deployment information to generate cooling demand characterization information for the one or more cabinets". This enables cooling control to associate the number, location, and expected load of cabinets, thereby introducing spatial and quantitative demand information at the starting point of the control logic. This provides the possibility of identifying and responding to differentiated cooling demands and helps alleviate the phenomenon of insufficient local cooling and excessive overall cooling caused by ignoring the distribution of heat sources in traditional solutions.

[0011] 2. This application achieves proactive matching calculation of supply capacity and demand information by performing the step of "matching and calculating the basic cooling capacity characterization information and the cooling demand characterization information to generate an initial cooling control strategy." This differs from the traditional approach of passively reacting only to temperature results. It can generate a more predictable and coordinated preliminary control scheme before the actual cooling demand is fully manifested in the rise of ambient temperature, thereby improving the initiative and initial energy efficiency of the control.

[0012] 3. This application incorporates a "feedback" correction mechanism based on environmental conditions into the "feedforward" control based on equipment and cabinet information by performing the step of "acquiring environmental status information collected by the pre-integrated monitoring and fire protection equipment module inside the equipment container, and modifying the initial cooling control strategy according to the environmental status information to generate a dynamically modified cooling control strategy." This combination enables the control strategy to possess both the planned nature based on system architecture and deployment, and the ability to respond to dynamic changes caused by factors such as climate and load fluctuations during actual operation. This enhances the adaptability of the control system to complex and variable operating environments, making the adjustment of cooling capacity both planned and flexible.

[0013] 4. This application implements the comprehensive strategy obtained from the aforementioned multi-source information processing and calculation by executing the step of "controlling the operating status of the cooling equipment module based on the dynamically modified cooling control strategy to control the cooling output of the prefabricated modular data center." This closed-loop process ensures that the cooling output is no longer a function of a single temperature signal, but rather a result of the combined dynamic balance of cooling equipment capacity, rack heat load demand, and real-time environmental conditions. Compared to traditional methods, this application improves the spatiotemporal matching between cooling supply and the actual heat load of IT equipment, thereby achieving relatively better overall cooling efficiency and energy utilization efficiency while ensuring the equipment's operating environment. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 A schematic diagram of a modular data center provided in this application embodiment; Figure 2 A cross-sectional structural diagram of a modular data center provided in an embodiment of this application; Figure 3 This is a schematic diagram of the overall process of a data center control method provided in an embodiment of this application; Figure 4 This is a hardware structure block diagram of an electronic device for a data center control method provided in an embodiment of this application.

[0016] Attached reference numerals: 1. Equipment container; 2. Cabinet; 3. Support device; 4. Low-voltage cable tray; 5. Secondary side ring pipe; 6. Row cabinet; 7. Air wall; 71. Refrigerant pipe; 72. Air conditioning distribution port. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0018] Reference Figure 1 This application also provides a modular data center, comprising: Equipment container 1, one or more cabinets 2, and support device 3 for low-voltage cable trays 4; The equipment container 1 is pre-integrated with a refrigeration equipment module, a power distribution equipment module, and a monitoring and fire protection equipment module. The one or more cabinets 2 are arranged independently of the equipment container 1 and are detachably connected to the refrigeration equipment module and power distribution equipment module inside the equipment container 1; The support device 3 is rotatably mounted on the top or side of the equipment container 1; The support device 3 has a retracted state and an extended state: in the retracted state, the support device 3 is rotated and retracted into the internal space of the equipment container 1 or close to the outer contour; in the extended state, the support device 3 is rotated out from the equipment container 1 and located above the one or more cabinets 2, for supporting and fixing the low-voltage cable tray 4.

[0019] Specifically, the core of a modular data center lies in the prefabrication and modular integration of data center infrastructure, decoupling it from IT cabinets 2 to achieve rapid deployment and flexible configuration. The data center mainly consists of three physical components: equipment container 1, one or more IT cabinets 2, and low-voltage cable trays 4 and supporting devices 3. The equipment container 1, as the core carrier of the infrastructure, completes most of the integration work in the factory; the IT cabinets 2, as load-bearing equipment, are independently deployed on-site; and the supporting devices 3 serve as auxiliary structures connecting the two, solving cable routing issues.

[0020] The interior of the container is planned into multiple functional areas and pre-integrated with the following three major modules: refrigeration equipment module, power distribution equipment module, and monitoring and fire protection equipment module.

[0021] The refrigeration equipment module integrates corresponding equipment according to the refrigeration scheme (air-cooled, liquid-cooled, or hybrid refrigeration) designed for the computer room. For the air-cooled scheme: the module mainly includes multiple precision air conditioning units. These air conditioning units are arranged along one or both sides of the container's length, and airflow is organized through common supply and return air ducts (i.e., air wall 7). Refrigerant pipes 71 and air conditioning power distribution ports 72 are located above them. Adjustable air valves are installed on the air ducts to distribute cooling capacity to different air supply areas. For the liquid-cooled scheme: the module mainly includes a liquid-cooled CDU, a manifold, and secondary side ring pipes 5 connecting them. The CDU is used to regulate the temperature, pressure, and flow rate of the coolant; the manifold serves as a distribution hub; and the quick-connect interface of the secondary side ring pipe 5 is led out to a dedicated panel on the container's side wall or end face. The hybrid refrigeration scheme is a combination of the above equipment. The controllers for all refrigeration equipment are connected to the integrated management system within the container.

[0022] Power Distribution Module: This module provides power to IT and refrigeration equipment. It mainly includes: Input Power Distribution Unit: Connects to external AC power or generator power. Uninterruptible Power Supply (UPS): Provides backup power for critical loads. Distribution Cabinet 6: Internally equipped with precision circuit breakers, smart meters (such as upstream of smart PDUs), and branch switches. Output Power Distribution Interface: Provides standardized power output interfaces via industrial connectors or dedicated socket panels. The power capacity design margin matches the power supply requirements of the integrated refrigeration equipment modules within the container.

[0023] The fire protection monitoring module is responsible for environmental sensing, safety early warning, and fire prevention. The monitoring component includes temperature and humidity sensors, leak detection ropes, and smoke detectors distributed in key locations within the container (such as supply / return air vents and near major equipment). It also integrates video surveillance cameras and access control systems. All sensor signals are connected to a local monitoring host. The fire protection component typically employs a gas extinguishing system (such as IG541 or heptafluoropropane). The cylinders and piping are prefabricated and installed in the factory, and detectors and nozzles are optimally arranged according to the internal space of the container.

[0024] The above three modules are installed, internally wired, tested for water and air circulation, and initially debugged in the factory, forming a fully functional "plug-and-play infrastructure" unit.

[0025] Rack 2 is a standard 19-inch or 21-inch network rack 2, or a full rack of servers. They are arranged independently of the equipment container 1; that is, rack 2 is not fixed inside the container, but deployed in the data center space outside the container. The arrangement of rack 2 follows the design principle of hot and cold aisle isolation.

[0026] The connection between cabinet 2 and equipment container 1 is detachable, specifically as follows: Connection with refrigeration equipment modules: Air-cooled scenario: Cabinet 2 is located in the cold aisle supplied by the container air conditioner, and is cooled by the airflow organization of the computer room space, without direct physical pipe connection. Liquid-cooled scenario: Each liquid-cooled cabinet 2 is connected to the corresponding secondary side ring pipe 5 quick-change interface on the side wall panel of the equipment container 1 through flexible quick-change pipes to realize the supply and return of coolant.

[0027] Connection with power distribution modules: Each cabinet 2 is connected to the corresponding output power distribution interface (socket) on the side wall panel of the equipment container 1 via a prefabricated cable (one end of which is a plug) to obtain power.

[0028] This detachable connection method allows the number, model, and location of rack 2 to be flexibly adjusted according to business needs, achieving decoupling of IT load and infrastructure.

[0029] The low-voltage cable tray 4 and the support device 3 are key structures for solving the problem of laying low-voltage (network, signal) cables between the cabinet 2 and the equipment container 1.

[0030] Structure and Installation: The main body of the support device 3 is one or a group of rigid trusses or brackets. Its root is rotatably mounted on the top (near the side edge) or side of the equipment container 1 via a heavy-duty swivel hinge. The axis of rotation is generally horizontal, allowing the support device 3 to rotate in the vertical plane.

[0031] Storage State: During the initial stages of transportation and container hoisting and positioning, the support device 3 is rotated to house its main structure within the reserved space inside the equipment container 1 (such as the top mezzanine), or rotated to fit snugly against the outer contour of the container (such as folding it to the side of the container and securing it with latches). In this state, the overall envelope size of the support device 3 does not exceed the standard transport dimensions of the container, ensuring transportation compliance.

[0032] Deployment Status: Once the equipment container 1 is in place in the computer room and the cabinets 2 are arranged, the operator releases the latches and rotates the support device 3 out from inside or to the side of the container. This rotation extends its truss or support arm to a position directly above one or more cabinets 2 arranged in a row. At this point, the support device 3 provides an elevated, stable installation path.

[0033] Functionality: In the unfolded state, the open or closed low-voltage cable tray 4 is installed and fixed on the support device 3. Subsequently, the low-voltage cables on the top of the cabinet 2 can be laid in the cable tray and finally introduced into the equipment container 1 to connect with network switches and other equipment. This avoids the complicated work of installing hangers and cable trays on the computer room ceiling, and realizes fast and neat low-voltage cabling.

[0034] The deployment and workflow of the data center can be as follows: Factory prefabrication: The integration and testing of all modules within equipment container 1 are completed in the factory.

[0035] Transportation and hoisting: The support device 3 is placed in its stowed state and transported as a whole to the data center site. The equipment container 1 is hoisted into the predetermined foundation position inside the computer room using a crane.

[0036] Positioning and Deployment: After the container is in place, connect its main interfaces for water, electricity, and network to the main system in the computer room. Then rotate the support device 3 to the deployed position and install and fix the low-voltage cable tray 4.

[0037] Rack 2 Deployment: Move the IT racks 2 to the designed location and arrange them in rows. Connect the power cables and liquid cooling pipes (if applicable) between each rack 2 and the equipment container 1.

[0038] Cable laying: Lay the network cable of cabinet 2 in the low-voltage cable tray 4 on the support device 3 to complete the connection.

[0039] Power-on test: Start up the infrastructure and IT equipment in sequence and perform joint debugging.

[0040] Reference Figure 3 This application provides a data center control method applicable to the aforementioned modular data center, comprising: Step 1: Obtain the configuration information of the refrigeration equipment modules pre-integrated in the equipment container, and parse the configuration information to generate basic refrigeration capacity characterization information of the equipment container; Step 2: Obtain the rack deployment information of one or more racks deployed on-site in the modular data center, and process the rack deployment information to generate cooling demand characterization information for the one or more racks; Step 3: Match and calculate the basic cooling capacity characterization information and the cooling demand characterization information to generate an initial cooling control strategy; Step 4: Obtain environmental status information collected by the monitoring and fire protection equipment module pre-integrated in the equipment container, and modify the initial cooling control strategy based on the environmental status information to generate a dynamically modified cooling control strategy; Step 5: Based on the dynamic correction cooling control strategy, control the operating status of the cooling equipment module to control the cooling output of the prefabricated modular data center.

[0041] The specific implementation process of Step 1 is as follows: First, a structured list of refrigeration equipment is obtained from the integrated management system of equipment containers in the prefabricated modular data center. This list serves as the configuration information for the refrigeration equipment modules. Next, this configuration information is parsed. This parsing process includes: identifying the type of each refrigeration sub-equipment in the list (e.g., distinguishing between air-cooled air conditioning units and liquid-cooled CDUs), recording its unique identifier, and extracting its rated cooling power parameters and physical installation location coordinates. Then, based on the parsed equipment type and rated cooling power parameters, the cooling capacity is aggregated and calculated separately for both air-cooled and liquid-cooled refrigeration paths: the rated cooling power of all air-cooled air conditioning units is added to obtain the total theoretical cooling capacity for the air-cooled path; the rated cooling power of all liquid-cooled CDUs is added to obtain the total theoretical cooling capacity for the liquid-cooled path. Finally, the parsed equipment identifiers, location coordinates, and the calculated total theoretical cooling capacity for both air-cooled and liquid-cooled paths are organized into a data structure with fixed fields. This data structure is named the basic cooling capacity characterization information. This information fully describes the types of refrigeration resources, spatial layout, and capacity limits that the equipment container can provide.

[0042] In the specific technical implementation of step 2: The control process first collects attribute data of each IT rack that is currently installed and powered on through a sensor network or data center infrastructure management system deployed on-site in the modular data center. This data constitutes the original rack deployment information. The processing of this rack deployment information begins with data cleaning and standardization, such as unifying power units. The core processing action is rack heat load modeling: based on the model identifier of each rack or directly read its internal server configuration, a preset rack power density mapping table is queried to assign a nominal heat load value to each rack. At the same time, based on the rack's position coordinates on the data center floor grid, a clustering algorithm (e.g., density-based spatial clustering algorithm) is used to spatially group all racks, dividing them into several physically adjacent rack clusters. Then, for each rack cluster, the nominal heat load values ​​of all racks within the cluster are summed to obtain the regional aggregated heat load of the cluster. Ultimately, the sum of the nominal heat load values ​​of all server racks is defined as the total heat load demand. The regional aggregated heat load of each server rack cluster is then associated with its corresponding physical area identifier (such as area number), and together they are encapsulated into a dataset called Cooling Demand Characterization Information. This information not only quantifies the total cooling demand but, more importantly, reveals the non-uniform distribution of heat load within the server room space.

[0043] In a specific scenario, step 3 is implemented as follows: The input to this step is the basic cooling capacity representation information and cooling demand representation information generated in the previous steps. The matching calculation first performs a cooling mode decision: comparing the total heat load demand in the cooling demand representation information with the total theoretical cooling capacity of the air-cooled and liquid-cooled paths in the basic cooling capacity representation information. The decision logic is: if the total heat load demand is lower than the air-cooled path capacity, the air-cooled mode is prioritized; if it is between the air-cooled and air-cooled plus liquid-cooled total capacity, the hybrid mode is activated; if it exceeds the total capacity, an alarm is triggered. This decision result is called the dominant cooling mode. Subsequently, the resource-demand mapping optimization stage begins. This stage constructs an optimization problem: its decision variables are the start / stop status and power operation percentage of each cooling device; its objective function is to minimize the overall cooling power consumption while satisfying the aggregated heat load (constraint) of all cabinet cluster areas; its constraints also include the upper and lower limits of the device's own power and the collaborative constraints between different cooling modes. Solving this optimization problem (using linear programming or heuristic algorithms) yields the specific target operating parameters (including start / stop and power setpoints) for each on-grid cooling device. Finally, the dominant cooling mode is integrated with the target operating parameters of all cooling equipment to form a complete and executable initial cooling control strategy.

[0044] The specific implementation process of step 4 is as follows: This step initiates a continuous data monitoring and processing loop. First, it receives raw data streams of environmental status information reported in real time from various sensors (such as temperature sensors, humidity sensors, and power meters) pre-integrated within the equipment container via the monitoring fire protection equipment module. After filtering (such as Kalman filtering) to remove noise, this raw data is converted into environmental status parameters with clear physical meaning, such as the real-time average temperature of different zones in the computer room, the intake / return air temperature of a specific cabinet, the cooling water return temperature, and the external temperature and humidity of the computer room. The core of the correction process is the incremental adjustment calculation of the strategy. This calculation compares the environmental status parameters with the set state expected to be achieved by the initial cooling control strategy (implied in the expected cooling output of the strategy) to generate a state deviation vector. A multi-input multi-output feedback controller (e.g., a decoupled PID controller array) receives this state deviation vector as input. In this feedback controller, for each controlled cooling zone or device, an independent control algorithm runs, and its output is a fine-tuning amount of the target operating parameters of the corresponding device in the initial cooling control strategy. For example, if the measured temperature in a certain area is higher than expected, the corresponding control loop outputs a positive power fine-tuning amount. Finally, the initial cooling control strategy is superimposed with all the calculated fine-tuning amounts to generate a dynamically corrected cooling control strategy that adapts to real-time operating conditions. This strategy inherits the global planning nature of the initial strategy while incorporating adaptive adjustments to instantaneous fluctuations.

[0045] In the specific technical implementation of step 5: the execution stage generates specific device-level control commands based on the dynamically modified refrigeration control strategy. First, control commands are encoded and distributed: the target operating parameters of each refrigeration device in the dynamically modified refrigeration control strategy are converted into standardized control command frames that can be recognized by the device controller using device-specific communication protocols (such as Modbus TCP, BACnet), and then distributed to the corresponding refrigeration device module execution units via the industrial network. Simultaneously, to form a closed loop, an execution status monitoring and strategy re-triggering process is initiated. This process continuously collects actual operational feedback from the refrigeration device modules (such as actual power consumption and actual flow rate) and the latest environmental status information obtained through monitoring fire protection equipment modules. This execution feedback information is used in two ways: firstly, directly for display and alarms on the monitoring interface; secondly, as a trigger condition or input for a new round of control. The system sets a strategy evaluation cycle or a key parameter deviation threshold. When the cycle is reached or the deviation exceeds the threshold, the current control cycle ends, and the entire process will automatically restart from step 2 or step 3 to obtain the latest cabinet deployment information or recalculate the strategy based on the latest feedback, thereby starting the next control cycle and ensuring that the cooling output continuously maintains a high degree of matching with the dynamically changing actual needs and environment.

[0046] Optionally, step 1, obtaining the configuration information of the pre-integrated refrigeration equipment modules within the equipment container and parsing the configuration information to generate basic refrigeration capacity characterization information for the equipment container, includes: Step 11: Extract the number and rated power of the air conditioning equipment and the number and rated power of the liquid-cooled CDU equipment included in the refrigeration equipment module from the configuration information; Step 12: Based on the number and rated power of the air conditioning equipment and the number and rated power of the liquid-cooled CDU equipment, calculate the maximum theoretical cooling capacity of the equipment container in air-cooled mode and liquid-cooled mode to generate the basic cooling capacity characterization information.

[0047] The specific implementation process of step 11 is as follows: This step begins by accessing the digital integrated archive of equipment containers in the prefabricated modular data center. This archive stores configuration information of refrigeration equipment modules recorded in structured data formats (such as JSON or XML). The specific parsing operation of this configuration information is performed by a dedicated configuration information parser. The parser first traverses the equipment list nodes of the configuration information, classifying and filtering the equipment entries in the list according to predefined equipment type identifiers (e.g., "AC" represents air-cooled air conditioning units, "L-CDU" represents liquid-cooled CDU equipment), thereby separating all air conditioning equipment entries and all liquid-cooled CDU equipment entries. Next, the parser extracts key parameters for each category of filtered equipment entries: for each air conditioning equipment entry, it extracts its "equipment quantity" field value (usually 1, representing an independent equipment unit) and its "rated power" field value (in kilowatts); for each liquid-cooled CDU equipment entry, it performs the same extraction operation to obtain its quantity and rated power. Finally, the parser summarizes the quantity and rated power of all extracted air conditioning units into an air conditioning unit parameter set, and summarizes the quantity and rated power of all liquid-cooled CDU units into a liquid-cooled CDU unit parameter set. These two sets together constitute the structured equipment capability parameters parsed from the original configuration information.

[0048] In a specific scenario, step 12 is implemented as follows: The input to this step is the structured equipment capability parameters obtained in the preceding steps, namely the set of air conditioning equipment parameters and the set of liquid-cooled CDU equipment parameters. The calculation process first processes the set of air conditioning equipment parameters using a weighted aggregation calculation method: the rated power value of each air conditioning unit in this set is multiplied by its corresponding number of units (usually 1), and then all the product results are summed. The sum obtained from this summation is initially defined as the total nominal power of the air conditioning equipment. Unlike the traditional simple addition, this application introduces a cooling capacity coefficient (this coefficient is a dimensionless value between 0 and 1, and its physical meaning reflects the reduction effect of the actual cooling capacity of multiple air conditioning units relative to the sum of their independent nominal powers in a prefabricated integrated environment due to factors such as airflow organization and mutual obstruction). Multiplying the total nominal power of the air conditioning equipment by the cooling capacity coefficient, the result is defined as the maximum theoretical cooling capacity of the equipment container in air-cooled mode. This calculation method introduces a reduction factor (i.e., a cooling capacity factor) reflecting the prefabricated integration scenario, making the assessed upper limit of cooling capacity closer to the actual achievable maximum value, thus providing a more reliable upper limit benchmark for subsequent capacity planning and matching. Similarly, the same weighted aggregation calculation and multiplication by the cooling capacity factor (the physical meaning of this factor may differ in the liquid cooling scenario, reflecting pipeline parallel losses, etc.) are performed on the liquid-cooled CDU equipment parameter set to obtain the maximum theoretical cooling capacity of the equipment container in liquid cooling mode. Finally, the calculated maximum theoretical cooling capacity in air-cooled mode and maximum theoretical cooling capacity in liquid-cooled mode, along with metadata such as equipment identifier and type parsed from the configuration information, are encapsulated into a unified data structure, which is named the basic cooling capacity characterization information, thereby completing the transformation from the original configuration to a quantitative capacity characterization.

[0049] The coefficient of performance (COP) is an empirical correction factor used to quantify the reduction in actual total cooling capacity relative to the sum of the individual rated power of each refrigeration unit when multiple refrigeration units operate simultaneously within the limited space and specific layout of a prefabricated container. Reasons for this reduction include: mutual airflow interference between units, return air short-circuiting, and limited space for condensation heat dissipation. This COP is not determined through inventive methods but can be obtained through conventional engineering procedures: during the container design finalization phase, all integrated refrigeration units are run in a simulated environment or on a real test platform, and their actual total cooling capacity under typical operating conditions is measured. This measured total cooling capacity is divided by the sum of the rated power of all units; the resulting ratio is the COP. This COP is typically less than 1 and may yield different values ​​for air-cooled and liquid-cooled systems.

[0050] For example: Suppose a container houses two air-cooled air conditioners, each with a rated cooling capacity of 50kW. The theoretical maximum cooling capacity, simply added together, is 100kW. However, in factory integration testing, when both air conditioners are running at full load, the actual measured total effective cooling capacity at the container outlet is only 92kW. Therefore, the "cooling capacity coefficient" for this container's air-cooled system is 92kW / 100kW = 0.92. In the calculation of step 12, the "maximum theoretical cooling capacity in air-cooled mode" used for evaluation will be 100kW × 0.92 = 92kW, not 100kW.

[0051] Optionally, step 2, obtaining the rack deployment information of one or more racks deployed on-site in the modular data center, and processing the rack deployment information to generate cooling demand characterization information for the one or more racks, includes: Step 21: Extract the total number of deployed cabinets, the power rating of each cabinet, and the location coordinates of each cabinet in the data center from the cabinet deployment information; Step 22: Calculate the total heat load requirement based on the total number of deployed cabinets and the power rating of each cabinet; Step 23: Decompose the total heat load demand into regional heat load demands corresponding to each air conditioning supply area or liquid cooling circuit partition in the refrigeration equipment module according to the location coordinates of each cabinet, so as to generate the refrigeration demand characterization information containing the total heat load demand and the regional heat load demand.

[0052] Optionally, obtaining the rack deployment information of one or more racks deployed on-site in the prefabricated modular data center includes: Step 24: Based on the total current business load planned for the prefabricated modular data center and the preset standard load capacity of a single rack, calculate the target number of racks required to meet the total current business load. Step 25: Generate the rack deployment information based on the target number of racks and the pre-set rack layout template of the prefabricated modular data center.

[0053] The specific implementation process of step 21 is as follows: This step begins by receiving raw data packets containing rack deployment information from a modular data center on-site monitoring system or a manual data entry system. These packets may contain unstructured logs or semi-structured tables. First, a data cleaning and verification module processes the raw data packets, including removing records with incorrect formats, standardizing coordinate units (e.g., converting all data packets to metric coordinates with a corner of the data center as the origin), and verifying the reasonableness of power ratings (e.g., limiting them to a preset reasonable power range). The standardized rack record list obtained after cleaning and verification serves as the input for the next extraction step. Next, a key field extractor traverses this standardized rack record list, performing three parallel extraction operations for each rack record in the list: reading and counting the "rack identifier" field to accumulate the total number of deployed racks; reading the "power rating" field and adding its value to a rack power rating set; and reading the "location coordinate X" and "location coordinate Y" fields and adding this pair of coordinates to a rack location coordinate set. Ultimately, the extractor outputs three distinct data entities: a scalar value representing the total number of deployed racks, a list of values ​​representing the set of rack power ratings, and a list of coordinate pairs representing the set of rack location coordinates. These three together constitute the core rack attribute parameters extracted from the original information.

[0054] In a specific scenario, step 22 is implemented as follows: This step uses the core rack attribute parameters obtained in step 21 as the processing object, specifically the total number of deployed racks and the set of rack power ratings. The process of calculating the total heat load demand is not simply adding up all power ratings. First, the set of rack power ratings is adjusted for instantaneous load. Considering the fluctuations in IT equipment load, this application introduces a dynamic load coefficient based on statistical history. This coefficient, for each rack or its type, is a value no greater than 1, generated over time or based on its load prediction model. Each rated power value in the set of rack power ratings is multiplied by the corresponding dynamic load coefficient to obtain a set of corrected instantaneous power values. Then, a summation operation is performed on this set of corrected instantaneous power values, and the sum is defined as the total heat load demand. This calculation method differs from the traditional approach of directly using the maximum power value on the nameplate. By introducing a dynamic coefficient reflecting the actual operating load level, the calculated total heat load demand is closer to the actual total heat that the rack group may generate at the present or near future, providing a more reasonable demand baseline for pre-matching cooling capacity.

[0055] The dynamic load factor is used to convert the rack's fixed power rating (nameplate value) into an estimate that more closely reflects its actual operating thermal load. It reflects the volatility of IT load. This factor is not arbitrarily set; its sources can be: a) Historical data analysis: monitoring the power consumption of this rack model or similar workloads over a historical period (e.g., 24 hours), calculating the ratio of its average power consumption to its rated power consumption, and using this as a base value for the factor. b) Preset values ​​based on load type: based on the type of server service deployed in the rack (e.g., web servers, databases, AI training), associating it with an industry-experienced typical load rate range, and taking the median or a value set according to current business priorities. For example, batch processing tasks may peak at night and be lower during the day. This factor is dynamically updated over time or according to task scheduling plans.

[0056] For example, consider an AI training server rack with a rated power of 10kW. According to its task schedule, the current period (e.g., morning) primarily involves data preprocessing, and historical data shows that the average power consumption during this phase is approximately 60% of the rated value. Therefore, at this moment, the rack's "dynamic load factor" can be set to 0.6. In step 22, this rack's contribution to the "total heat load demand" is 10kW × 0.6 = 6kW, not 10kW.

[0057] In the specific technical implementation of step 23: The inputs to this step include the total heat load demand calculated in step 22, the set of cabinet location coordinates extracted in step 21, and a pre-set cooling equipment coverage area mapping table. This table defines the air supply area boundary of each air conditioner or the collaborative air supply area of ​​multiple air conditioners, as well as the physical cabinet grouping corresponding to each liquid cooling loop. The core of the processing is spatial load allocation. First, based on the set of cabinet location coordinates, a coordinate-area matching algorithm based on a pre-set area grid is used to map each cabinet coordinate to a specific area defined in the cooling equipment coverage area mapping table (e.g., "Column A Air Conditioner Air Supply Area - 1" or "Liquid Cooling Loop - 3"). This mapping process assigns a home area identifier to each cabinet. Next, regional demand aggregation is performed: all cabinets that have been assigned home area identifiers are traversed. For all cabinets mapped to the same home area identifier, their corresponding corrected instantaneous power values ​​(from the calculation process in step 22) are accumulated. This accumulated value is defined as the regional heat load demand of that area. Finally, the total heat load demand and all calculated regional heat load demands (each demand associated with its corresponding regional identifier) ​​are encapsulated to form a hierarchical data volume, which is named the Cooling Demand Characterization Information. This information not only includes the overall demand, but more importantly, reveals the detailed distribution of heat load within the computer room space based on the coverage of cooling equipment capacity.

[0058] The coordinate-region matching algorithm is based on a predefined region grid mapping table. The server room floor is pre-divided into regular grids (e.g., 1m × 1m), each grid associated with a "region identifier," which corresponds to a specific air conditioning supply area or liquid cooling circuit zone. The algorithm converts the rack coordinates (x, y) into the index of its corresponding grid, and then queries the mapping table to obtain its assigned region identifier. The grid division is based on the actual physical coverage of the cooling equipment, which can be determined through design drawings or on-site testing. For example, the server room is divided into four 2.5m × 2.5m grids, labeled Zone_A, Zone_B, Zone_C, and Zone_D. The coordinates of a rack are measured as (1.2m, 1.8m). The algorithm determines that these coordinates fall within the lower left corner grid (range: 0≤x<2.5, 0≤y<2.5), therefore its assigned region identifier is Zone_A.

[0059] Clustering algorithms are used to automatically group nearby server racks when there is no pre-defined grid or dynamic grouping is required. Density-based spatial clustering algorithms (such as DBSCAN) can be used. The core parameter "neighborhood radius (eps)" is typically set slightly larger than the typical rack spacing (e.g., if the rack spacing is 1 meter, then eps can be set to 1.5 meters), and the parameter "minimum number of samples (min_samples)" can be set to 1, so that each rack can be a seed for a cluster. The rack clusters generated by the clustering results can be directly regarded as a cooling area or associated with the nearest cooling equipment. For example: there are 5 racks with coordinates (1,1), (1,2), (1,3), (5,5), (5,6). Setting eps=1.5 and min_samples=1, the algorithm clusters the first three racks within 1.5 meters of each other into Cluster1, and the last two into Cluster2. These two clusters can be regarded as two independent cooling demand areas.

[0060] The preferred implementation process of step 24 is as follows: This optional step is executed during the planning phase. Its processing objects are the current total workload of the business system (e.g., represented by the number of virtual machines, total computing tasks, or total storage capacity) and a predefined single-rack standard load capacity lookup table. First, the current total workload is normalized, that is, based on the type of workload, it is converted into an equivalent number of "standard rack load units" using the lookup table, resulting in a normalized total workload unit quantity. Then, the rack quantity is calculated: the normalized workload unit quantity is divided by the single-rack standard load capacity (i.e., the number of standard load units that each standard rack can support). The quotient obtained from this division may be a non-integer. This application adopts a rounding-up and redundancy-addition strategy here: first, the quotient is rounded up to obtain a basic integer number of racks; then, based on a preset business growth redundancy coefficient (e.g., reserving a certain percentage of capacity for the future), this basic integer number of racks is multiplied by (1 + business growth redundancy coefficient), and finally, the result is rounded up again to obtain the final target number of racks. This strategy differs from simple division and rounding. By introducing redundancy considerations for the inconvenience of expansion in prefabricated modular scenarios, the calculated number of racks can not only meet the current load, but also reserve physical space for near-term growth, making the generated deployment plan more forward-looking.

[0061] The business growth redundancy factor is a strategic parameter used in the planning phase to reserve physical space and infrastructure capacity (power, cooling) for foreseeable future business growth when calculating the required number of server racks, thus avoiding frequent expansions. This factor is determined based on business planning; for example, if business volume is planned to grow by 20% in the next 6 months, this factor could be set to 0.2. It is a standard engineering planning reservation method, and its value is set by the data center operator based on business development forecasts and a trade-off between expansion costs.

[0062] For example: Based on the current business load, at least 50 standard server racks are needed. Considering the company's business development plan, the load is expected to increase by 25% within the next year. The operations decision-makers set a "business growth redundancy coefficient" of 0.25 for this plan. Therefore, the "target number of server racks" calculated in step 24 is based on 50 racks. After considering redundancy, it becomes 50 × (1 + 0.25) = 62.5, which, rounded up, is 63 racks. The generated "rack deployment information" will reserve space for these 63 racks.

[0063] A rack layout template is a structured database or configuration file containing the following fields: Location ID, X coordinate, Y coordinate, Rated power (kW), Cold aisle number, Bound cooling zone ID, and Status (Available / Occupied). This template is generated during the data center design phase and reflects the capacity and relationships of the physical infrastructure.

[0064] The layout rules are a set of heuristic rules for automatically selecting rack locations, which can be executed in order of priority: Continuity rule: Prioritize the selection of consecutive available locations within the same cold aisle to form a complete cold / hot aisle isolation.

[0065] Load balancing rule: Among multiple cold aisles, the aisle with the lower total power of currently deployed racks is selected to balance the power load.

[0066] Cooling association rules: Prioritize locations that are close to or directly associated with the target cooling equipment (based on business load characteristics).

[0067] For example: Suppose the template has cold aisle A (locations A1-A5) and cold aisle B (locations B1-B5), and 3 racks need to be deployed. Two locations in aisle A are already used, and one location in aisle B is already used. According to the continuity rule, the system will prioritize selecting consecutive locations B2, B3, and B4 in aisle B (assuming they are available), rather than selecting locations across aisles or non-consecutive locations.

[0068] In a specific scenario, step 25 is implemented as follows: The input to this step is the target number of server racks calculated in step 24 and the pre-existing prefabricated modular data center rack layout template within the system. This layout template is a data structure that defines the coordinates, power capacity, and cooling resource allocation (air zone or liquid cooling loop) of all available server rack locations within the data center, and may include layout rules such as "first fill the same cold aisle" and "power zoning balance". The process of generating rack deployment information is an automatic layout generation process based on templates and constraints. First, based on the target number of server racks, server rack locations that meet the power and cooling capacity requirements and are in the "available" state are selected from the rack layout template to form a candidate location list. Then, the layout rules defined in the template are applied to sort and select the locations in the candidate location list. For example, using the rule of "close arrangement to optimize cold aisle closure" will prioritize physically contiguous locations. One design element of this application is the introduction of a cooling efficiency-priority evaluation: for multiple candidate location subsets that meet basic rules, the expected cooling efficiency of each subset layout is estimated (e.g., based on the ratio of existing load to cooling capacity of the cooling zone to which the location belongs), and the subset with the higher expected cooling efficiency is selected as the final layout. Finally, for each selected rack location, a record is generated containing a virtual "rack identifier," the location's inherent "power rating" (taken from a template), and "location coordinates." The set of all these records is then used to generate the rack deployment information. This method transforms deployment planning from manual design to an automated process based on rules and efficiency evaluation, suitable for the rapid deployment needs of prefabricated modular data centers.

[0069] The cooling efficiency evaluation is based on the regional cooling capacity utilization rate and airflow organization score. Regional cooling capacity utilization rate: For each candidate rack layout subset, the expected additional heat load (estimated based on rack rated power and dynamic load factor) for each cooling zone where the new rack will be placed is calculated. Then, the ratio of the total additional heat load to the cooling capacity (i.e., utilization rate) of the zone is calculated. The subset that results in the most balanced utilization rate (minimum variance) across all zones and is below a safe threshold (e.g., 85%) is selected.

[0070] Airflow organization score: assesses the contribution of subset layout to hot and cold aisle isolation. For example, a layout that completely fills a cold aisle scores higher than a dispersed layout.

[0071] The final evaluation can be a weighted average of the two: Overall score = α (1-Utilization variance)+β Airflow organization score.

[0072] For example, assuming there are 3 target cabinets, each with a rated power of 20kW and a dynamic load factor of 0.8 for the current business period, the expected heat load for each cabinet is 16kW. Cooling zone A has a cooling capacity of 100kW and an existing heat load of 50kW, while cooling zone B has a cooling capacity of 80kW and an existing heat load of 30kW. For candidate location subset X, if all three cabinets are placed in the location bound to cooling zone A, the cooling capacity utilization rate after deployment in zone A is (50 + 16 × 3) / 100 = 98%, exceeding the safety threshold of 0.85. Therefore, candidate location subset X is excluded. For candidate location subset Y, if two cabinets are located in cooling zone A and one cabinet is located in cooling zone B, then the cooling capacity utilization rate of zone A is (50+16×2) / 100=82%, and the cooling capacity utilization rate of zone B is (30+16) / 80=57.5%. Both are below the safety threshold of 0.85, therefore candidate location subset Y can enter the subsequent comprehensive evaluation. For other candidate location subsets that also meet the safety threshold, their cooling capacity utilization rate dispersion and airflow organization score are compared. Finally, the candidate location subset that neither causes excessive local cooling zone capacity nor fails to maintain good hot and cold aisle isolation continuity is selected as the target location subset.

[0073] Optionally, step 3, matching and calculating the basic cooling capacity characterization information and the cooling demand characterization information to generate an initial cooling control strategy, includes: Step 31: Compare the total heat load demand in the cooling demand characterization information with the maximum theoretical cooling capacity in the basic cooling capacity characterization information to determine whether the activated cooling mode is air-cooled mode, liquid-cooled mode or hybrid mode. Step 32: If it is determined that the air-cooled mode or the hybrid mode is enabled, then according to the regional heat load demand in the cooling demand characterization information, calculate and allocate the number of start-ups, operating power and air supply direction of each air conditioner in the cooling equipment module, so as to generate the first part of the control instructions related to the air conditioner in the initial cooling control strategy. Step 33: If it is determined that liquid cooling mode or hybrid mode is enabled, then according to the regional heat load demand in the cooling demand characterization information, calculate and allocate the number of liquid-cooled CDU devices to be started, the operating power and the corresponding secondary side loop flow rate in the cooling equipment module, so as to generate the second part of the control instructions related to the liquid-cooled equipment in the initial cooling control strategy.

[0074] In a specific scenario, step 31 involves processing the total heat load demand from the cooling demand characterization information obtained in the preceding steps, as well as the maximum theoretical cooling capacity in air-cooled mode and liquid-cooled mode from the basic cooling capacity characterization information. The primary step in the matching calculation is the cooling mode decision. This decision is not a simple binary comparison but involves a multi-level judgment logic. First, the total heat load demand is compared with the maximum theoretical cooling capacity in air-cooled mode. If the total heat load demand is less than or equal to a preset percentage of the maximum theoretical cooling capacity in air-cooled mode (for example, this percentage could be set to 80%, intended to reserve a certain adjustment margin for the air-cooled system and consider its partial load efficiency), it is preliminarily determined that the conditions for enabling pure air-cooled mode are met. Next, to assess the necessity of the hybrid mode, the total heat load demand is compared with the sum of the maximum theoretical cooling capacity in air-cooled mode and the maximum theoretical cooling capacity in liquid-cooled mode. If the total heat load demand exceeds the sum of the two, it is determined that the cooling capacity is insufficient and an alarm is triggered. If the total heat load demand is between the two, further evaluation is required. This application introduces a cooling efficiency and switching cost evaluation function here. This function comprehensively considers the overall energy consumption estimated based on the equipment's partial load efficiency curve, under the current total heat load demand, whether using air cooling alone, liquid cooling alone (if capacity is sufficient), or a combination of both. It also considers the potential equipment wear and tear and stabilization time costs associated with switching from one mode to another. The function calculates the comprehensive "cost" of various potential modes and selects the mode with the lowest comprehensive cost as the final dominant cooling mode (i.e., air cooling, liquid cooling, or a hybrid mode). This decision-making mechanism differs from traditional single-threshold comparisons. By introducing a multi-objective evaluation of efficiency and switching costs, it ensures that mode selection, while meeting cooling requirements, prioritizes the option with better overall energy efficiency and more stable operation.

[0075] The cooling efficiency and switching cost evaluation function is a decision function used to choose among multiple cooling modes (air cooling, liquid cooling, hybrid). It quantifies the "overall cost" of different options. The "cooling efficiency" part can be estimated by consulting the partial load efficiency curves provided by the equipment manufacturer, which will estimate the power consumption expected to be generated by operating in a certain mode under the current total heat load demand. The "switching cost" part may include: the startup time cost of the equipment from sleep to full load (converted into the possible high temperature risk during this period), the wear and tear on the equipment life due to frequent start-ups and shutdowns of the compressor or pump (converted into maintenance costs), and the risk cost of system instability during mode switching. The function can be a weighted sum of these component costs, i.e.: Overall cost = α × Estimated energy consumption + β × Time cost + γ × Equipment wear and tear cost. The weights α, β, and γ are determined by the operating strategy (e.g., if energy saving is prioritized, α has a higher weight). The overall cost of each candidate mode is calculated, and the one with the lowest cost is selected.

[0076] For example: Assume the current total heat load demand is 80kW, the maximum capacity of the air-cooled system is 100kW, and the maximum capacity of the liquid-cooled system is 60kW. Option A (pure air cooling): Estimated energy consumption is a certain value. Option B (pure liquid cooling): Insufficient capacity, not included in the evaluation. Option C (hybrid mode: air cooling handles 60kW, liquid cooling handles 20kW): Estimated energy consumption is another value. Furthermore, if the current system is in air-cooled mode, switching to hybrid mode requires the liquid cooling system to start, with an estimated 5-minute stabilization time. The function will calculate the "overall cost" (mainly energy cost) of Option A and the "overall cost" (energy cost + risk cost due to the 5-minute switching time) of Option C. If the overall cost of Option C is lower, the hybrid mode will be selected.

[0077] The specific implementation process of step 32 is as follows: When step 31 determines that the air-cooled mode or the hybrid mode is enabled, this step is initiated to generate air conditioning control commands. The core inputs for processing are the regional heat load demand in the cooling demand characterization information (each demand is associated with its assigned region identifier) ​​and the location and capacity data of the air conditioning equipment in the basic cooling capacity characterization information. First, air conditioning equipment-region association mapping is performed. Based on the preset "Air Conditioning Air Supply Area Coverage Table", one or more regions (i.e., its set of responsible regions) that each air conditioning unit is responsible for cooling are determined. Next, demand-capacity matching and power allocation are performed. For each air conditioning unit, its set of responsible regions is traversed, and the regional heat load demand of these regions is summed to obtain the aggregate heat load that the air conditioning unit needs to bear. Then, the aggregate heat load of the air conditioning unit is compared with its rated power. If the aggregate heat load is zero, a "shutdown" command is generated; if the aggregate heat load is greater than zero and less than its rated power, a target operating power percentage (aggregate heat load / rated power) is calculated. If the aggregated heat load reaches or exceeds the rated power, the target operating power percentage is set to 100%, and a neighboring air conditioner coordination alarm may be triggered. This calculation process is performed one by one for each air conditioner that needs to be activated in all cooling modes to obtain the target operating power percentage for each device. In addition, for air conditioners with adjustable air supply direction, this application adopts an air supply angle calculation method based on regional demand weight. According to the regional heat load demand of each region in its responsibility area set, a demand weight is assigned to each region, and then a weighted average air supply vector is calculated based on these weights, which is then converted into a specific target air supply angle. Finally, the identifiers of all air conditioners that need to be activated, their calculated target operating power percentages, and the target air supply angle (if adjustable) are encapsulated into a set of executable instructions. This set of instructions is named the first part of the control instructions related to the air conditioners in the initial cooling control strategy.

[0078] Among them, the air supply angle calculation based on regional demand weight is achieved through vector calculation and angle transformation.

[0079] Direction vector definition: For air conditioning equipment, its location is taken as the origin O. Each responsibility area i has a representative point Pi (usually the geometric center of the area), and the vector pointing from O to Pi is the direction vector of that area. Weight assignment: The weight Wi is directly taken as the weighted average vector of the area's heat load demand (or the normalized value): V_avg=(Σ(Wi) V_i)) / ΣWi, where V_i is the direction vector of region i (which can be converted to a unit vector). Angle conversion: Calculate the angle between V_avg and the reference direction (such as true north or the direction of the cabinet arrangement), which is the target air supply angle. If the air conditioner only supports a finite number of discrete angles, then take the closest discrete value.

[0080] For example: One air conditioner serves two areas: Area 1 has a heat load demand of 30kW and a directional angle of 0° (directly in front); Area 2 has a heat load demand of 10kW and a directional angle of 90° (to the right). Then the weighted average directional angle = (30... 0°+10 (90°) / (30+10)=22.5°. The air conditioner's airflow angle is adjusted to 22.5°.

[0081] In the preferred technical implementation of step 33: when step 31 determines to enable liquid cooling mode or hybrid mode, this step is executed in parallel to generate liquid cooling control commands. Its processing objects are also the regional heat load demand in the cooling demand characterization information, and the data regarding liquid-cooled CDU equipment and secondary-side loop circuits in the basic cooling capacity characterization information. First, liquid cooling loop-cabinet attribution analysis is performed. Based on the predefined "Liquid Cooling Secondary-Side Loop Circuit Allocation Table," the specific cabinet list served by each liquid cooling loop (usually associated with one or a group of liquid-cooled CDU equipment) is clarified, thereby establishing the attribution relationship between liquid cooling loops and cabinets. Next, loop heat load summarization and CDU load allocation are performed. For each liquid cooling loop, based on the attribution relationship between the liquid cooling loop and cabinets, all cabinets belonging to that loop are found, and the regional heat load demand (representing the heat load of these cabinets) of the area where these cabinets are located is summed to obtain the total demand heat load of that liquid cooling loop. Then, based on the connection relationship between the liquid-cooled CDU equipment and the loop (e.g., one CDU may pump multiple loops), the total demand heat load of each loop is allocated to its corresponding liquid-cooled CDU equipment, forming the total CDU load that each CDU equipment needs to handle. This application, in calculating the target operating power and target flow rate, differs from simple proportional scaling by introducing a flow-power coordinated model based on the loop temperature difference setting. This model uses the total demand heat load of the liquid-cooled loop and the preset target inlet and outlet temperature difference of the coolant as a basis to back-calculate the minimum theoretical volumetric flow rate required to meet the heat load exchange. Simultaneously, based on the total CDU load and the CDU equipment performance curve, its target operating power is determined. Finally, an instruction containing the target operating power is generated for each liquid-cooled CDU equipment to be activated, and a regulating valve instruction containing the target flow rate is generated for each controlled secondary-side loop. The set of these instructions is named the second part of the control instructions related to the liquid-cooled equipment in the initial refrigeration control strategy. This model, through coordinated control of flow rate and power, aims to ensure that the liquid-cooled system meets cooling requirements while maintaining the energy consumption of the delivery pump at a relatively low level.

[0082] Among them, the flow-power coordinated model based on the loop temperature difference setting is a fundamental thermodynamic relationship application model for liquid cooling system control. Its core principle is: Q = c ρ V ΔT, where Q is the heat load to be dissipated (i.e., the total required heat load), c is the specific heat capacity of the coolant, ρ is the density of the coolant, V is the volumetric flow rate (i.e., the target flow rate), and ΔT is the temperature rise of the coolant after passing through the cold plate (i.e., the preset target inlet and outlet temperature difference). Given Q, c, ρ, and ΔT, the required minimum theoretical volumetric flow rate V_min = Q / (c ρ (ΔT). Meanwhile, the power of the CDU pump is related to the flow rate and system resistance. The target operating power can be found from the target flow rate V based on the pump's performance curve. The model's "synergy" is reflected in: prioritizing the determination of the minimum flow rate required to achieve the cooling effect based on the heat load and temperature difference, and then determining the pump power accordingly, rather than setting the pump power first and then adjusting the flow rate, thus theoretically prioritizing the minimization of delivery energy consumption. c represents the product calculation. ρ represents the product between c and ρ.

[0083] For example, a liquid cooling circuit needs to dissipate a heat load (Q) of 30kW. The coolant is a 50% ethylene glycol aqueous solution, and its c and ρ are known physical properties. The target temperature difference ΔT is set to 10°C (e.g., inlet 20°C, outlet 30°C). Substituting these values ​​into the formula, the required minimum theoretical flow rate V_min can be calculated, for example, 0.5 L / s. The control system sends this 0.5 L / s as the target flow rate for the circuit to the regulating valve. Simultaneously, according to the CDU pump curve, delivering 0.5 L / s may require the pump to operate at 40% power; this 40% is taken as the target operating power of the CDU.

[0084] Optionally, step 4, obtaining environmental status information collected by the pre-integrated monitoring and fire protection equipment module inside the equipment container, and modifying the initial cooling control strategy based on the environmental status information to generate a dynamically modified cooling control strategy, includes: Step 41: Extract the external ambient temperature of the computer room, the real-time temperature of different monitoring points inside the computer room, and the real-time power consumption data of the cabinet fed back by the power distribution equipment module from the environmental status information. Step 42: Based on the comparison between the external ambient temperature of the computer room and the preset temperature threshold, adjust the operating mode setting of the air conditioning equipment in the initial cooling control strategy; Step 43: Based on the deviation between the real-time temperature of different monitoring points inside the computer room and the heat load demand of the corresponding area in the cooling demand characterization information, perform proportional-integral-derivative adjustment on the operating power of the air conditioning equipment in the corresponding area or the flow rate of the corresponding liquid cooling circuit in the initial cooling control strategy. Step 44: Based on the difference between the real-time power consumption data of the cabinet and the corresponding cabinet power rating in the cabinet deployment information, dynamically update the regional heat load demand in the cooling demand characterization information, and recalculate the initial cooling control strategy based on the updated regional heat load demand; Step 45: The adjusted, proportional-integral-derivative control, and recalculated control commands are fused together to generate the dynamic correction refrigeration control strategy.

[0085] In a specific scenario, step 41 involves processing the raw data stream continuously reported by various sensors monitoring the fire protection equipment module within the equipment container. This data stream represents the environmental status information. First, a multi-source data synchronization and preprocessing unit receives this data stream, performs timestamp alignment, removes anomalous jump values ​​(e.g., using sliding window mean filtering), and standardizes data units, outputting a set of synchronized and cleaned sensor readings. Next, a key physical quantity extraction engine analyzes these readings. Based on a predefined sensor identifier and physical quantity mapping table, the engine identifies and separates three types of core data from synchronized and cleaned sensor readings: temperature values ​​from readings identified as "outdoor temperature and humidity sensors" are extracted as the external ambient temperature of the data center; readings identified as "indoor temperature sensors" are grouped according to their deployment location coordinates to form a real-time temperature mapping table for internal monitoring points of the data center, with the monitoring point location identifier as the key and the real-time temperature value as the value; readings identified as "rack smart PDUs (Power Distribution Units)" or "power meters" are analyzed to retrieve their real-time power values, and combined with their associated rack identifiers to form a rack real-time power consumption data mapping table, with the rack identifier as the key and the real-time power consumption value as the value. Finally, the external ambient temperature of the data center, the real-time temperature mapping table for internal monitoring points, and the rack real-time power consumption data mapping table are encapsulated into a structured set of environmental state parameters that can be used for subsequent correction calculations.

[0086] The specific implementation process of step 42 is as follows: This step takes the external ambient temperature of the computer room obtained in step 41 and the first part of the control instructions related to the air conditioning equipment in the initial cooling control strategy as the processing objects. The core of the adjustment process is the external climate adaptive mode switching. First, the external ambient temperature of the computer room is compared with a preset natural cooling activation temperature threshold. If the external ambient temperature of the computer room is lower than the threshold, a suggestion to enable the natural cooling mode is generated. The core of this application here is that it does not directly force the switch, but performs a mode switching feasibility assessment. This assessment checks whether the operating status of the air conditioning equipment, refrigerant pressure, and air valve mechanism in the current first part of the control instructions support the natural cooling mode, and predicts the stability of the internal temperature field of the computer room after the switch. If the assessment passes, a set of specific air conditioning equipment operating mode modification instructions are generated, for example, changing the operating mode of a specific air conditioner from "mechanical cooling" to "hybrid mode" or "full natural cooling mode", and adjusting its target fan speed accordingly. If the external ambient temperature of the computer room is higher than another preset high temperature efficiency optimization threshold, an adjustment suggestion to optimize mechanical cooling efficiency is generated, such as adjusting the condenser fan control logic or enabling the energy-saving operation curve. These air conditioning equipment operation mode modification instructions constitute the first revision of the first part of the control instructions, generating the initially revised air conditioning control instructions.

[0087] In the preferred technical implementation of step 43: This step simultaneously processes the real-time temperature mapping table of monitoring points inside the computer room from the structured environmental state parameter set in step 41, the cooling demand characterization information calculated in steps 22 and 23 (especially the regional heat load demand associated with the area identifier), and the initial cooling control strategy from steps 32 and 33 (including the first part of the control instructions and the second part of the control instructions). The core of the correction is a refined proportional-integral-derivative adjustment based on spatial temperature feedback. First, a monitoring point-control area association is performed: based on the location coordinates of each monitoring point, its corresponding cooling control area (i.e., an air conditioning supply area or a liquid cooling loop service area) is determined, thereby mapping the readings in the real-time temperature mapping table of monitoring points inside the computer room to the actual measured temperature of the corresponding area. Next, the regional temperature deviation is calculated for each control area, which is the difference between the actual measured temperature of the area (take the average or highest value of all monitoring points in the area) and the preset target temperature setpoint of the area (this setpoint can be dynamically fine-tuned according to the regional heat load demand of the area). Then, a set of independent proportional-integral-derivative controllers are activated, with each controller responsible for one control area. Each controller takes the regional temperature deviation as input. The adjustment in this application lies in the dynamic tuning of the controller parameters: based on the current regional heat load demand level (from cooling demand characterization information), the proportional gain, integral time, and derivative time of the controller corresponding to that region are dynamically adjusted. For example, for regions with higher heat loads, a more aggressive proportional gain is used to obtain a faster response. The controller output is a fine-tuning of the equipment control parameters serving that region in the initial cooling control strategy: for air-cooled regions, the output is an incremental adjustment to the percentage of the target operating power of the air conditioner; for liquid-cooled regions, the output is an incremental adjustment to the target flow rate of the secondary loop pipe. These fine-tuning values ​​are applied to the initially corrected air conditioning control command and the second part of the control command to generate a corrected command adjusted by proportional-integral-derivative (PI-DE) regulation.

[0088] Dynamic tuning of controller parameters is an adaptive optimization of the classic proportional-integral-derivative (PID) controller. The parameters (proportional coefficient Kp, integral time Ti, and derivative time Td) of a conventional PID controller are fixed values. "Dynamic tuning" refers to adjusting these parameters based on the real-time state of the controlled object (in this case, the regional heat load demand). Specifically, when the regional heat load demand is high, it means the heat source intensity is high, and the temperature rises rapidly, requiring a faster controller response and stronger ability to suppress deviations. In this case, the proportional coefficient Kp can be automatically increased, and the integral time Ti may be decreased (to enhance the integral action and eliminate steady-state error). Conversely, when the heat load is low, Kp can be decreased to prevent over-adjustment and oscillations. The adjustment rules can be based on a set of predefined empirical tables or simple linear relationships describing the correspondence between "heat load level" and "recommended controller parameter set".

[0089] For example, a region responsible for cooling a high-performance computing cluster typically has a heat load demand exceeding 50kW. Its proportional-integral-derivative (PID) controller is pre-configured with the parameter set: Kp_high (larger value), Ti_low (smaller value). Another region responsible for storage servers has a stable heat load demand of around 10kW, and its pre-configured parameter set is: Kp_low, Ti_high. During system operation, the system directly calls the corresponding parameter set based on the real-time heat load demand of each region. If the demand in the high-load region suddenly increases from 50kW to 70kW, the system still uses the Kp_high, Ti_low parameter set; if the demand in the low-load region remains at 10kW, it uses Kp_low, Ti_high. This differs from using a fixed set of compromise parameters that are not optimal in any situation.

[0090] The specific technical implementation process of step 44 is as follows: This step processes the cabinet real-time power consumption data mapping table obtained in step 41 from the structured environmental state parameter set, the core cabinet attribute parameters extracted in step 21 (especially the set of cabinet power ratings), and the original cooling demand characterization information. This step realizes online calibration of the demand model. First, power consumption deviation calculation and filtering are performed: For each cabinet in the cabinet real-time power consumption data mapping table, its corresponding rated power in the set of cabinet power ratings is found, and the ratio of real-time power consumption to rated power is calculated to obtain the instantaneous load rate of the cabinet. To avoid the influence of instantaneous fluctuations, a short-time moving average filter is applied to the instantaneous load rate of each cabinet to obtain a smoothed load rate. Next, regional heat load demand re-estimation is performed: Based on the cabinet-region affiliation relationship (from the mapping in step 23), the smoothed load rate of all cabinets in a region is multiplied by their respective rated power, and then these products are summed to obtain the corrected regional heat load demand based on real-time measurements. This corrected regional heat load demand may be significantly different from the initial regional heat load demand calculated based on nominal values ​​or prediction coefficients. Then, a strategy recalculation process is triggered: the corrected regional heat load demand for all regions is newly calculated, replacing the corresponding part in the original cooling demand characterization information to form updated cooling demand characterization information. Using the updated cooling demand characterization information and the basic cooling capacity characterization information as input, a matching calculation process similar to steps 31 to 33 is re-executed, but this calculation can inherit the current equipment operating state as the initial condition, thereby generating a set of candidate control strategies recalculated based on real-time power consumption.

[0091] The smoothed load rate is the result of filtering the directly measured instantaneous load rate. The purpose is to eliminate the interference caused by instantaneous spikes or brief troughs in heat load estimation, obtaining a load rate value that better represents the stable trend. The "short-time moving average filtering" used is a conventional data processing method. For example, it takes the instantaneous load rate values ​​of the most recent 10 sampling periods (assuming once per second, i.e., the most recent 10 seconds), calculates their arithmetic mean, and uses this as the smoothed load rate for the current moment. This avoids misjudgments of heat load demand caused by instantaneous power consumption fluctuations, and prevents frequent operation of cooling equipment as a result.

[0092] For example, the instantaneous load rate sampling values ​​of a certain server rack over a continuous 10 seconds are: [58%, 60%, 85%, 61%, 59%, 60%, 62%, 88%, 59%, 60%]. There are two abnormally high points of 85% and 88%, which may be due to instantaneous computational tasks. Using a moving average (window size 10), the smoothed load rate at the current moment = (58+60+85+61+59+60+62+88+59+60) / 10 = 65.2%. The system uses this relatively stable value of 65.2% to update the thermal load demand, instead of using the last instantaneous value of 60% or the highest value of 88%, thus achieving a more robust correction.

[0093] In a specific scenario, step 45 is the final step in the fusion and decision-making of correction instructions. Its processing objects include: the air conditioning equipment operation mode modification instruction generated in step 42, the correction instruction generated in step 43 after proportional-integral-derivative adjustment, and the candidate control strategy generated in step 44 based on real-time power consumption recalculation. The fusion process first performs instruction conflict detection and resolution. For example, the air conditioning equipment operation mode modification instruction may require switching to natural cooling mode, while the correction instruction after proportional-integral-derivative adjustment requires increasing the compressor power of the same air conditioner; these two conflict. This application employs a priority-weighted fusion algorithm. This algorithm assigns priority weights to different types of correction instructions. For example, mode switching instructions based on safety or energy efficiency (from step 42) have higher priority, fine-tuning instructions based on real-time temperature feedback (from step 43) have medium priority, and candidate strategies based on periodic recalculation (from step 44) only have high priority when the change exceeds a certain threshold. The algorithm iterates through all controlled devices. For each control parameter (such as power and mode) of each device, it collects suggested values ​​for that parameter from all commands to be fused. Based on their priority, it calculates a weighted average or selects the highest priority command value to obtain the final fused setpoint for that parameter. Finally, the final fused setpoints of all devices are organized into a complete, expendable command set according to device type and communication protocol requirements. This command set is named the Dynamic Correction Cooling Control Strategy. This strategy integrates multi-source corrections such as climate adaptation, real-time feedback, and demand model updates, aiming to ensure that the control output maintains a long-term energy efficiency optimization trend while responding promptly to short-term dynamic disturbances.

[0094] In the priority-weighted fusion algorithm, the weight allocation principles include: Safety / mode instructions (such as switching to natural cooling): highest weight (e.g., w=1.0), usually with veto power. Real-time feedback fine-tuning instructions (proportional-integral-derivative output): medium weight (e.g., w=0.6~0.8), reflecting immediate adjustment needs. Periodic recalculation instructions: low weight (e.g., w=0.3~0.5), representing medium- to long-term optimization suggestions.

[0095] The conflict resolution process includes: collecting all instruction values ​​and weights for the same parameter on the same device. If an instruction has an absolutely high weight (e.g., a threshold of 0.9 or higher), that value is used directly. Otherwise, a weighted average is calculated: final value = Σ(instruction value_i) Weight_i) / Σ(weight_i). Perform a reasonableness check on the weighted average result (e.g., does not exceed equipment limits).

[0096] For example, an air conditioner receives three instructions regarding its target power parameters: A (mode switching requirement set to 50%, weight 1.0), B (proportional-integral-derivative fine-tuning +5%, weight 0.7), and C (recalculation suggestion 60%, weight 0.4). Since A has a weight of 1.0, 50% is directly adopted, and B and C are ignored.

[0097] Optionally, step 5, controlling the operating status of the cooling equipment module based on the dynamically corrected cooling control strategy to control the cooling output of the prefabricated modular data center, includes: Step 51: According to the dynamic correction cooling control strategy, send a control signal containing the operating mode, target power and air supply angle to the air conditioning equipment in the cooling equipment module, and send a control signal containing the target power and target flow rate to the liquid-cooled CDU equipment in the cooling equipment module. Step 52: According to the dynamic correction cooling control strategy, adjust the on / off state and power distribution of the power supply circuit associated with the cooling equipment module in the power distribution equipment module, and start the monitoring and fire protection equipment module to continuously monitor the adjusted cooling output effect and the computer room environmental parameters, so as to complete the closed-loop control of the cooling output of the prefabricated modular data center.

[0098] In a specific scenario, step 51 begins with parsing and distributing instructions to the dynamically modified cooling control strategy generated in step 45. First, a control instruction encoder receives the dynamically modified cooling control strategy as input. Based on the different device types (air conditioning units or liquid-cooled CDU units) and their communication protocol specifications within the strategy, the encoder converts abstract setpoints into specific, standardized device control instruction frames recognizable by the device controller. For air conditioning units, the standardized device control instruction frame includes the operating mode (e.g., "mechanical cooling," "natural cooling"), the target power setpoint (expressed as a percentage or absolute power value), and the airflow angle setpoint (in degrees). For liquid-cooled CDU (Cooling Distribution Unit) units, the standardized device control instruction frame includes the target power setpoint and the target flow rate setpoint for its associated secondary side loop pipe. One design feature of this application is the introduction of instruction security and validity pre-verification. Before transmission, the system checks the rationality of the instruction, for example, whether the target power exceeds the device's safety range, whether the airflow angle is within mechanical limits, and whether the combination of mode and power conforms to the device's operating procedures. Verified commands are marked as deployable command frames. Subsequently, a multiplexed communication scheduler, based on the device's network address and communication protocol (such as Modbus TCP / IP, BACnet), sends the deployable command frames to the corresponding air conditioning equipment controller and liquid-cooled CDU equipment controller in the refrigeration equipment module via the appropriate industrial network channel. This process differs from traditional one-way transmission; pre-verification and scheduling ensure the reliability of control commands and the safety of the equipment.

[0099] In the pre-verification of command security and validity, the verification rules are stored in a device capability and safety rule base. Each rule is in the form of "IF (condition) THEN (action: pass / reject / alarm)". Rule sources include: device technical manuals (power range, angle range, mode compatibility table); system integration specifications (e.g., minimum interval for mode switching, minimum number of parallel devices); and safety procedures (e.g., prohibiting the shutdown of all cooling equipment at high temperatures). During verification, the parameters in the standardized device control command frame are compared against the rule base one by one. Only when all parameters pass the comparison can the frame be marked as ready to be issued.

[0100] For example, a rule in the rule base states: "IF Air conditioner operating mode is 'natural cooling' THEN Compressor power must be 0". If a command requires the mode to be natural cooling but the compressor power is set to 30%, the verification will fail, the command will be rejected, and an alarm will be triggered.

[0101] The specific technical implementation process of step 52 is as follows: This step realizes cross-subsystem coordination and closed-loop monitoring. Its processing object is also the dynamic correction of the refrigeration control strategy, as this strategy implicitly contains the power configuration information required by the refrigeration equipment. First, power supply strategy parsing and power distribution command generation are performed. Based on the start / stop and power settings of the air conditioning equipment and liquid-cooled CDU equipment in the dynamic correction of the refrigeration control strategy, their total power demand and branch circuit demand are calculated. Subsequently, a power distribution coordination controller sends control commands to the power distribution equipment module. These commands include: connecting or disconnecting specific power supply circuits that supply power to the refrigeration equipment about to be started, and adjusting the power distribution setpoints on the relevant circuits (e.g., adjusting the output current limit via a smart PDU). The key to this application here is the introduction of refrigeration-power distribution coordinated interlocking logic. This logic ensures that before performing on / off operations on the power supply circuit, it checks whether the refrigeration equipment is in a safe state that allows power outage (e.g., the compressor has stopped, the pump has stopped). Conversely, before starting high-power refrigeration equipment, it confirms that its power supply circuit is ready and has sufficient capacity, thereby avoiding surge impacts or equipment damage. While adjusting the power distribution, the system starts a high-frequency monitoring cycle. This cyclic command monitors the sensor network in the fire protection equipment module, continuously collecting indirect characterization parameters of the adjusted cooling output effect (such as the air outlet temperature of each air conditioner, the supply and return water temperature difference of the liquid cooling circuit, and the temperature change rate of each area) and comprehensive data center environmental parameters (such as temperature, humidity, and airflow velocity) at a frequency higher than the control cycle. This real-time collected data is encapsulated into a closed-loop feedback data packet. This data packet is pushed to the decision logic entry point of the control system in real time. The system sets explicit cyclic trigger conditions; for example, when a key parameter in the closed-loop feedback data packet deviates from its expected range to a certain threshold, or when a fixed cycle has elapsed since the last strategy generation, the current control cycle ends, and the entire process automatically restarts from step 2 (obtaining the latest cabinet deployment information) or step 4 (starting a new correction calculation based on the latest closed-loop feedback data packet). This signifies the completion of the current closed loop and the start of a new one, thus achieving continuous, adaptive closed-loop control of the cooling output of the prefabricated modular data center.

[0102] The query mechanism for state checks and the calculation method for capacity checks in the refrigeration-power distribution coordinated interlocking logic can be implemented as follows: Status check query: Before performing power distribution operations (on / off), read the current operating status word of the refrigeration equipment through the equipment monitoring interface (such as IPMI, Modbus) to determine whether it is in a state that allows power off, such as "standby", "stop" or "fault".

[0103] Capacity check calculation: Before starting the cooling equipment, query the current real-time load power P_current (read from the smart PDU) and the rated capacity P_rated of the power distribution circuit. Calculate the estimated total load after starting the equipment = P_current + P_device_startup (peak power at equipment startup, which can be conservatively taken as the rated power). Requirement: Estimated total load ≤ P_rated Safety factor (e.g., 0.9).

[0104] Interlock Execution: Only if all the above checks pass can the power distribution operation or equipment start command be executed; otherwise, the process will be interrupted and an alarm will be triggered.

[0105] For example: We plan to start an air conditioner with a rated power of 50kW. First, we check its status; it is "off," indicating that starting is permitted. Then, we check its power supply circuit: the current load is 80kW, the rated capacity is 150kW, and the safety factor is 0.9. The calculation is: 80 + 50 = 130kW < 150kW. 0.9 = 135kW, capacity check passed. Lockout logic released, start command allowed.

[0106] Optionally, in one embodiment, in step 25, based on the target number of racks and the pre-set rack layout template of the prefabricated modular data center, rack deployment information is generated, including: Step 251: Select available rack locations from the rack layout template that meet the power capacity constraints and the cooling capacity constraints of the corresponding cooling area, and construct multiple candidate location subsets based on the target number of racks; Step 252: For each candidate location subset, based on the power rating of the cabinet to be deployed and the dynamic load coefficient representing the corresponding service load level, determine the expected new heat load generated by the candidate location subset in each cooling zone, and based on the expected new heat load, the existing heat load of each cooling zone and the cooling capacity of each cooling zone, determine the cooling capacity utilization rate of each cooling zone after the cabinet is deployed. Step 253: Determine the dispersion of cooling capacity utilization rate corresponding to candidate location subsets based on the cooling capacity utilization rate of each cooling zone, and determine the airflow organization score based on the influence of candidate location subsets on the continuity of hot and cold aisle isolation. Step 254: From the candidate location subset where the cooling capacity utilization rate of each cooling zone is lower than the preset safety threshold, determine the target location subset based on the dispersion of cooling capacity utilization rate and airflow organization score, and generate cabinet deployment information based on the target location subset.

[0107] In the specific implementation of step 251, the rack layout template corresponding to the prefabricated modular data center is first read. The rack layout template establishes location records according to the pre-planned rack installation positions within the data center. Each location record includes at least a location ID, location coordinates, the rated power allowed for rack configuration at that location, the cold aisle number, the bound cooling zone ID, and the location status. The location status indicates whether the corresponding rack location is currently available for rack deployment, and the bound cooling zone ID determines which air conditioning supply zone or liquid cooling circuit zone should handle the heat load generated after rack deployment at that location. Locations with a status of occupied, reserved, or unavailable are not included in the candidate range for this rack deployment; for locations with a status of available, further checks on power capacity constraints and cooling capacity constraints are performed.

[0108] Power capacity constraints can be determined based on the power rating of the rack to be deployed and the rated power registered in the rack layout template for the corresponding location. Specifically, when selecting a location for a rack, the power rating of the rack and the rated power of the location to be matched are read. If the power rating of the rack is not greater than the rated power of the location, the location is determined to meet the power capacity constraint corresponding to the rack. If the power rating of the rack is greater than the rated power of the location, it means that the location cannot provide the power capacity required to meet the rack's rated operating needs, and the location is removed from the available locations corresponding to the rack. For multiple racks to be deployed with different power ratings, locations can be matched sequentially from highest to lowest power rating to avoid locations with higher rated power being occupied by racks with lower power requirements, leaving no available locations for high-power racks.

[0109] For cooling capacity constraints, the cooling zone to which the location belongs is determined based on the bound cooling zone ID of the location to be matched, and the current existing heat load and corresponding cooling capacity of the cooling zone are read. The existing heat load can be determined based on the heat load requirements of the cabinets already deployed in the cooling zone, and the corresponding cooling capacity can be determined based on the cooling capacity of the air conditioning equipment or liquid-cooled CDU equipment associated with the cooling zone. In the screening stage, the power rating of the cabinet to be deployed is used as the upper limit estimate of the new heat load. When the corresponding cooling zone still has a cooling margin to support the cabinet to be deployed after deducting the existing heat load, the location is determined to pass the preliminary check of cooling capacity constraints. This check is used to exclude locations that obviously cannot meet the cooling requirements before forming a subset of candidate locations. The subsequent step 252 further calculates the actual expected heat load generated after multiple cabinets are deployed simultaneously based on the dynamic load coefficient.

[0110] After completing the above screening, a set of available locations corresponding to each rack to be deployed is obtained. Based on the target number of racks obtained in step 24, a number of rack locations equal to the target number are selected from the set of available locations, and a one-to-one correspondence is established between the racks to be deployed and the selected locations, thus forming a subset of candidate locations. When constructing the subset of candidate locations, only one rack to be deployed is allowed to correspond to the same rack location. At the same time, the cold aisle numbers in the rack layout template can be used to preferentially form a subset of candidate locations composed of consecutive locations within the same cold aisle, while other subsets of candidate locations formed by combinations of available locations in different cold aisles are retained for subsequent comparison of cooling capacity utilization and airflow organization. This results in multiple subsets of candidate locations that all meet the basic deployment conditions but have different rack space distribution methods.

[0111] In the specific implementation of step 252, heat load prediction is performed on each candidate location subset obtained in step 251. For each rack to be deployed in the candidate location subset, the expected heat load of the rack is determined based on its power rating and corresponding dynamic load factor. The meaning and determination method of the dynamic load factor are consistent with those in step 22, reflecting the actual load level of the corresponding rack during the current time period or the scheduled service operation period. For the i-th rack to be deployed, its power rating is denoted as Pi, and its dynamic load factor as ki. The expected heat load corresponding to this rack can then be determined as Pi × ki. Since the dynamic load factor is a value no greater than 1, the obtained expected heat load is no greater than the corresponding rack's power rating, thus avoiding the situation where the cooling demand estimate is too high due to directly using the full-load operation of all racks as the actual planned heat load.

[0112] After determining the expected heat load of each rack to be deployed, the expected heat load of each rack in the candidate location subset is assigned to its corresponding cooling zone based on the bound cooling zone ID corresponding to each rack location. If multiple racks exist within the same cooling zone, their expected heat loads are summed, and the sum is used as the expected additional heat load generated by that candidate location subset in that cooling zone. Thus, for each candidate location subset, a set of expected additional heat loads corresponding to a cooling zone ID can be formed.

[0113] Subsequently, the existing heat load of the corresponding cooling area before the deployment of the racks to be deployed is read. The existing heat load can be determined by aggregating the corrected real-time power values ​​of each rack already deployed in the corresponding cooling area, or it can be directly used from the regional heat load demand of the corresponding cooling area in the current cooling demand characterization information. The existing heat load of a cooling area is added to the expected new heat load generated by the candidate location subset in the cooling area to obtain the expected total heat load of the cooling area after the rack deployment is completed according to the candidate location subset.

[0114] For the j-th cooling zone, let its existing heat load be Qj0, the expected new heat load generated by the candidate location subset be ΔQj, and the cooling capacity of the corresponding cooling zone be Cj. Then, the cooling capacity utilization rate Uj after deployment can be determined according to Uj=(Qj0+ΔQj) / Cj. The cooling capacity Cj is the effective cooling capacity that the cooling zone can provide under the current cooling method. For air-cooled zones, it can be determined based on the cooling capacity of the air conditioning equipment responsible for the zone; for liquid-cooled zones, it can be determined based on the cooling capacity of the liquid-cooled CDU equipment responsible for the corresponding secondary side loop pipe. By calculating each cooling zone participating in this deployment evaluation in the same way, a set of cooling capacity utilization rates corresponding to each candidate location subset can be obtained. This set of cooling capacity utilization rates can characterize whether the candidate location subset will cause a certain cooling zone to approach or exceed the upper limit of cooling capacity, and can also serve as the basis for subsequent calculations to determine whether the distribution of cooling resources in each zone is balanced.

[0115] In the specific implementation of step 253, the spatial balance of cooling load caused by each candidate location subset and the impact of rack arrangement on the airflow organization of hot and cold aisles are evaluated respectively. Regarding the spatial balance of cooling load, under the same candidate location subset, the cooling capacity utilization rate corresponding to each cooling area participating in the deployment evaluation is obtained, and the average value of these cooling capacity utilization rates is first determined. Let the number of cooling areas participating in the evaluation be m, and the cooling capacity utilization rates corresponding to each cooling area be U1 to Um, respectively. The average utilization rate is the sum of the cooling capacity utilization rates divided by m. Further, the square of the deviation between each cooling capacity utilization rate and the average utilization rate is calculated, and the average of the squared deviations is obtained to obtain the variance of the cooling capacity utilization rate corresponding to the candidate location subset. This variance is used as the dispersion of the cooling capacity utilization rate. The smaller the variance, the closer the capacity utilization status of multiple cooling areas is, and the lower the degree of high load in one cooling area and low load in other cooling areas caused by centralized rack deployment; the larger the variance, the more obvious the difference in capacity utilization status between different cooling areas.

[0116] For airflow organization scoring, based on the cold aisle numbers in the rack layout template and the arrangement order of each location within its corresponding cold aisle, the locations in the candidate location subset are merged with the already occupied locations in the data center to obtain the expected rack arrangement after deployment according to the candidate location subset. For each cold aisle, the continuity of occupied rack locations is determined according to the arrangement order of the locations. An optional evaluation method is to identify the continuous location segment with the most continuously occupied rack locations within each cold aisle and calculate the ratio between the number of rack locations contained in this continuous location segment and the total number of valid rack locations in the cold aisle, as the continuity of the cold aisle; further, the airflow organization score of the candidate location subset is determined based on the continuity of each cold aisle involved in this deployment. Candidate location subsets that can fill the gaps between existing racks and form a continuous arrangement of existing and new racks have relatively high airflow organization scores; candidate location subsets that disperse new racks across multiple cold aisles or form multiple separate rack segments within the same cold aisle have relatively low airflow organization scores.

[0117] When evaluating using the above method, the airflow organization score directly reflects the impact of the candidate layout on the continuity of cold and hot aisle isolation. When racks are arranged continuously within the same cold aisle, it reduces the bypass paths of hot and cold air formed by gaps between racks, which is beneficial for maintaining a relatively complete spatial separation between the cold and hot aisles. When racks are arranged dispersedly, it is easy to form multiple unoccupied intervals. Therefore, step 253 does not only evaluate the layout based on the geometric distance between racks, but also simultaneously obtains the dispersion of cooling capacity utilization, which characterizes the spatial balance of cooling resources, and the airflow organization score, which characterizes the cold and hot aisle isolation state, as two evaluation parameters for subsequently selecting the final deployment location.

[0118] In the specific implementation of step 254, a cooling capacity safety check is first performed on each candidate location subset. A safety threshold corresponding to the cooling capacity utilization rate is preset. This safety threshold is less than the utilization rate corresponding to the maximum cooling capacity of the cooling area, in order to reserve a certain dynamic adjustment margin for the cooling equipment. For example, the safety threshold can be set to 0.85, that is, after the expected completion of rack deployment, the expected total heat load of a cooling area will not exceed 85% of its effective cooling capacity. For a candidate location subset, if the cooling capacity utilization rate of any cooling area reaches or exceeds the safety threshold, the candidate location subset is excluded; only when the cooling capacity utilization rate of each cooling area corresponding to the candidate location subset is lower than the safety threshold is it retained as a location subset to be evaluated. If there is no candidate location subset that meets the safety threshold requirement, it means that under the current target number of racks and existing cooling resources, a rack deployment scheme that meets the preset cooling margin requirement cannot be formed. At this time, a prompt message indicating insufficient cooling capacity can be output, and the generation of corresponding rack deployment information can be stopped, or the candidate location screening can be re-executed.

[0119] For each candidate location subset that passes the security check, a comprehensive evaluation is performed based on the dispersion of cooling capacity utilization and the airflow organization score calculated in step 253. In one implementation, the cooling capacity utilization can be expressed as a ratio between 0 and 1, with the variance of the cooling capacity utilization as the dispersion D and the airflow organization score F as the evaluation value for the continuity of the hot and cold aisles. The comprehensive score S is determined according to a preset first weight α and a second weight β, where S = α × (1 - D) + β × F, where α and β are both non-negative and α + β = 1. The first weight is used to characterize the influence of the cooling capacity balance on the cabinet layout evaluation, and the second weight is used to characterize the influence of the airflow organization state on the cabinet layout evaluation. A comprehensive score is calculated for each location subset to be evaluated, and the candidate location subset with the highest comprehensive score is determined as the target location subset. When there are multiple candidate location subsets with the same comprehensive score, the highest cooling capacity utilization corresponding to each candidate location subset can be further compared, and the candidate location subset with the lower highest cooling capacity utilization is selected first to retain a relatively larger local cooling adjustment margin.

[0120] After determining the target location subset, rack deployment information is generated based on the correspondence between the racks to be deployed and their locations established within the target location subset. For each rack to be deployed, at least one rack deployment record is generated, containing a rack identifier, power rating, and location coordinates. The rack identifier distinguishes different racks to be deployed, the power rating uses the rated power corresponding to that rack, and the location coordinates use the location coordinates corresponding to that rack within the target location subset. The cold aisle number and bound cooling zone ID of the corresponding location can also be associated and saved with the rack deployment record for later use in determining the corresponding air conditioning supply zone or liquid cooling circuit zone based on the rack location. All rack deployment records corresponding to the racks to be deployed together form the rack deployment information and serve as the data basis for subsequent calculations of total heat load demand and regional heat load demand.

[0121] For example, assuming there are 3 target cabinets, each with a rated power of 20kW and a dynamic load factor of 0.8 for the current business period, the expected heat load for each cabinet is 16kW. Cooling zone A has a cooling capacity of 100kW and an existing heat load of 50kW, while cooling zone B has a cooling capacity of 80kW and an existing heat load of 30kW. For candidate location subset X, if all three cabinets are placed in the location bound to cooling zone A, the cooling capacity utilization rate after deployment in zone A is (50 + 16 × 3) / 100 = 98%, exceeding the safety threshold of 0.85. Therefore, candidate location subset X is excluded. For candidate location subset Y, if two cabinets are located in cooling zone A and one cabinet is located in cooling zone B, then the cooling capacity utilization rate of zone A is (50+16×2) / 100=82%, and the cooling capacity utilization rate of zone B is (30+16) / 80=57.5%. Both are below the safety threshold of 0.85, therefore candidate location subset Y can enter the subsequent comprehensive evaluation. For other candidate location subsets that also meet the safety threshold, their cooling capacity utilization rate dispersion and airflow organization score are compared. Finally, the candidate location subset that neither causes excessive local cooling zone capacity nor fails to maintain good hot and cold aisle isolation continuity is selected as the target location subset.

[0122] Optionally, in one embodiment, step 52, adjusting the on / off state and power distribution of the power supply circuit associated with the cooling equipment module in the power distribution equipment module, includes: Step 521: Before disconnecting the power supply circuit associated with the target refrigeration equipment to be powered off, obtain the current operating status of the target refrigeration equipment, and release the power-off lockout for the power supply circuit when the current operating status belongs to the preset power-off allowed state. Step 522: Before starting the target refrigeration equipment to be started, obtain the current real-time load power, rated capacity of the power supply circuit associated with the target refrigeration equipment, and the starting peak power of the target refrigeration equipment. Determine the estimated total load after starting based on the current real-time load power and the starting peak power, and release the start-up lockout for the target refrigeration equipment when the estimated total load is not greater than the product of the rated capacity of the circuit and the preset safety factor. Step 523: After the corresponding interlock is released, execute the on / off control of the power supply circuit or the start control of the target refrigeration equipment, and acquire closed-loop feedback data characterizing the cooling output effect at a monitoring frequency higher than the refrigeration control cycle. The closed-loop feedback data includes at least two of the following: air conditioner outlet temperature, liquid cooling circuit supply and return water temperature difference, and temperature change rate of each refrigeration zone. Step 524: When the key parameters in the closed-loop feedback data deviate from the corresponding expected range to the preset deviation threshold, trigger the acquisition of rack deployment information again, or recalculate the dynamic correction cooling control strategy based on the latest closed-loop feedback data.

[0123] In the specific implementation of step 521, when the dynamic correction of the refrigeration control strategy indicates the shutdown of a certain air conditioning unit or liquid-cooled CDU unit, and it is necessary to further disconnect the power supply circuit corresponding to the refrigeration unit, a circuit breaker command is not directly sent to the power distribution equipment module. Instead, the corresponding equipment monitoring interface and power supply circuit identifier are first determined based on the equipment identifier of the target refrigeration unit, and the current operating status of the target refrigeration unit is read through the equipment monitoring interface. The equipment monitoring interface can be a status reading interface provided by the refrigeration unit's own controller, or it can read the operating status word of the target refrigeration unit from the integrated management system through communication methods such as Modbus. The operating status word is used to characterize that the target refrigeration unit is currently in one of the following states: running, stopped, standby, starting, shutdown transition, or fault.

[0124] Pre-establish permissible power-off states for different types of refrigeration equipment in the equipment capability and safety rule base. For air conditioning equipment, a permissible power-off state is defined at least after the compressor has stopped running and the equipment has entered a shutdown or standby state; for liquid-cooled CDU equipment, a permissible power-off state is defined at least after the circulating pump has stopped and the equipment has entered a shutdown or standby state. In cases where equipment is in a fault state, it can be determined whether the fault state falls under the permissible direct power-off state according to the corresponding equipment's safety rules, thus avoiding treating all fault states uniformly as permissible power-off conditions. Existing manuals already clearly use operating status words to identify "standby," "shutdown," or fault states that conform to the rules.

[0125] The control system matches the current operating status it reads with the permissible power-off status of the target refrigeration equipment. If the current operating status matches the permissible power-off status, the power-off interlock status set for that power supply circuit is switched from the interlocked state to the unlocked state, allowing the power distribution module to accept subsequent power supply circuit disconnection commands. If the current operating status is still in the running state, start-up state, or other states where power-off is not permitted, the power-off interlock remains, and disconnection commands are not sent to the corresponding power supply circuit. For the target refrigeration equipment that is still in the running state, the shutdown control corresponding to step 51 can continue to be executed, and the power-off interlock is released only after the reread operating status meets the permissible power-off conditions. In this way, the mechanical operating status of the refrigeration equipment itself becomes a prerequisite for the power distribution circuit power-off operation, preventing the compressor or circulating pump from losing power before completing the shutdown process.

[0126] In the specific implementation of step 522, when the dynamic correction cooling control strategy determines that a certain air conditioning unit or liquid-cooled CDU unit needs to be started, the pre-established correspondence between the unit and the power supply circuit is first queried based on the unit identifier of the target cooling unit to obtain the target power supply circuit for powering the target cooling unit. Subsequently, the real-time load power of the power supply circuit is read through the intelligent PDU, power meter, or other power detection units in the power distribution equipment module corresponding to the target power supply circuit, and the rated capacity of the power supply circuit is read from the power distribution configuration data.

[0127] The peak startup power of the target refrigeration equipment can be obtained from the equipment configuration data or technical parameters. When the peak startup power is not recorded separately, the rated power of the target refrigeration equipment can be used as the peak startup power for capacity checking in a relatively conservative manner. Let the current real-time load power of the target power supply circuit be P_current, and the peak startup power of the target refrigeration equipment be P_start. Then, the estimated total load P_est after the target refrigeration equipment starts up can be determined as the sum of P_current and P_start.

[0128] Simultaneously, a safety factor K less than or equal to 1 is set for the target power supply circuit. The upper limit of the allowable safe power P_safe is determined by multiplying the circuit's rated capacity P_rated by the safety factor K, i.e., P_safe = P_rated × K. The safety factor is used to ensure that the power supply circuit retains a certain capacity margin after the equipment starts up; for example, it can be set to 0.9. When P_est is not greater than P_safe, it indicates that the power supply circuit is still within the preset safe capacity range after considering the starting power of the target cooling equipment, and the control system releases the start-up interlock set for the target cooling equipment. When P_est is greater than P_safe, the start-up interlock remains in place, the start-up command for the target cooling equipment is not executed, and an alarm message indicating insufficient power supply circuit capacity is generated.

[0129] For example, if the peak starting power of the target refrigeration equipment is 50kW, and the current real-time load power of its corresponding power supply circuit is 80kW, the rated capacity is 150kW, and the safety factor is 0.9, then the estimated total load is 130kW, and the upper limit of the safe power is 135kW. Therefore, the capacity check passes and the start-up lockout is released.

[0130] In the specific implementation of step 523, the control sequence is adopted according to whether the current process is a power-off process or a startup process. When the power-off process is executed, after the power-off interlock is released in step 521, the power distribution coordination controller sends a disconnect command to the power supply circuit associated with the target refrigeration equipment, causing the corresponding branch switch to switch to the disconnected state. When the startup process is executed, after the startup interlock is released in step 522, it is first confirmed that the target power supply circuit is in the connected state. If the target power supply circuit is not yet connected, a circuit connection command is first sent to the power distribution equipment module. After confirming that the corresponding power supply circuit is in the power supply state, a startup command is then sent to the target refrigeration equipment. Thus, "power-off is allowed after equipment stops" and "start is allowed after power supply capacity is sufficient" form two control constraints in different directions.

[0131] While performing the aforementioned power supply circuit switching or target cooling equipment startup operations, a monitoring process for the current control results is initiated. Let the control cycle of the computer room cooling control strategy be T_c. The monitoring and fire protection equipment module collects relevant operating parameters at a sampling cycle T_m, which is less than T_c, even if the monitoring frequency is higher than the update frequency of the cooling control strategy. In this way, multiple sets of cooling effect feedback data can be obtained between two cooling strategy updates, instead of waiting for the next control cycle to arrive before judging the actual effect of the current power distribution and cooling equipment operation. The existing manual also clearly requires that this monitoring cycle collect data at a frequency higher than the control cycle.

[0132] For air conditioning equipment, the air outlet temperature is obtained through a temperature sensor located at the air outlet. For liquid cooling equipment, the supply and return liquid temperatures are obtained through temperature detection points located on the supply and return sides of the secondary loop pipe, respectively, and the difference between the return and supply liquid temperatures is defined as the supply and return water temperature difference of the liquid cooling loop. The rate of temperature change for each cooling zone can be determined by the ratio of the difference in zone temperatures obtained at two consecutive sampling times to the time interval between the two sampling times. The zone temperature can be the average of the real-time temperatures of multiple monitoring points within the zone, or the highest temperature can be used according to actual control requirements, so that the rate of temperature change can characterize the trend of temperature rise or fall in the corresponding zone after the refrigeration equipment is adjusted.

[0133] For the air conditioning outlet temperature, liquid cooling circuit supply and return water temperature difference, and area temperature change rate obtained within the same monitoring period, these parameters can be correlated according to equipment identification, cooling area identification, and sampling time, and at least two of these parameters should be selected to form closed-loop feedback data. For cooling areas using only air cooling, the air conditioning outlet temperature and area temperature change rate can be mainly used; for liquid cooling areas, the liquid cooling circuit supply and return water temperature difference and area temperature change rate can be mainly used; for mixed cooling areas using both air and liquid cooling, all three types of parameters can be collected simultaneously. This avoids requiring different cooling structures to obtain feedback parameters that are not applicable to their own structures, while ensuring that the closed-loop feedback data reflects the actual cooling output and changes in the area's thermal state after the cooling equipment is activated.

[0134] In the specific implementation of step 524, corresponding expected ranges and deviation thresholds are set for various parameters in the closed-loop feedback data. The expected ranges can be determined based on the desired operating state of the current dynamically adjusted cooling control strategy and the normal operating range of the cooling equipment. For example, expected temperature ranges can be set for the air conditioner outlet temperature, the liquid cooling circuit supply and return water temperature difference, and the cooling zone temperature change rate. For each set of closed-loop feedback data, it is determined whether the key parameters are within their corresponding expected ranges.

[0135] If the key parameters are within the expected range, the current dynamic correction cooling control strategy continues and monitoring is performed. If the key parameters exceed the expected range, the deviation of the actual parameters from the expected range boundary is further calculated. For example, when the measured value is higher than the upper limit of the expected range, the difference between the measured value and the upper limit can be used as the deviation; when the measured value is lower than the lower limit of the expected range, the difference between the lower limit and the measured value can be used as the deviation. When the deviation reaches the corresponding preset deviation threshold, it is determined that a deviation has occurred between the current cooling output effect and the expected state of the dynamic correction cooling control strategy, requiring readjustment, and a new control loop is triggered.

[0136] After triggering a new control loop, it's essential to first determine whether the currently saved rack deployment information still matches the actual rack deployment status in the modular data center. Specifically, this can be done by rereading the rack identifiers, quantities, and location coordinates of currently deployed and powered-on racks through the data center infrastructure management system or on-site monitoring network, and comparing them with the corresponding data in the current rack deployment information. If the number of racks, rack identifiers, or location coordinates change, step 2 is re-executed to regenerate the cooling demand characterization information based on the latest on-site deployment status. Furthermore, a new initial cooling control strategy and a dynamically corrected cooling control strategy are then generated based on this.

[0137] If the rack deployment information is confirmed to remain unchanged, there is no need to re-establish the rack deployment information. Instead, the latest closed-loop feedback data is used as part of the current environmental status information, and the process re-enters the correction process in step 4. Specifically, the current temperature status, liquid cooling loop operating status, and related cooling effect parameters of each cooling zone can be re-determined. Combined with existing cooling demand characterization information, the operating power of the corresponding air conditioning equipment, the target flow rate of the liquid cooling loop, or other current control parameters are re-corrected. Then, an updated dynamic correction cooling control strategy is formed according to the aforementioned instruction fusion method. The updated dynamic correction cooling control strategy then re-enters step 5 for execution, thus completing a closed-loop control process consisting of "strategy execution—power distribution and equipment interlocking—high-frequency effect monitoring—deviation judgment—strategy regeneration".

[0138] The methods and embodiments provided in this application can be executed in electronic devices such as mobile terminals, computer terminals, servers, or similar computing devices. Figure 4 This is a hardware structure block diagram of an electronic device for a data center control method provided in an embodiment of this application. (See diagram below.) Figure 4 As shown, the electronic device 900 can vary significantly due to differences in configuration or performance. It may include one or more central processing units (CPUs) 910 (CPUs 910 may include, but are not limited to, microprocessors such as MCUs or programmable logic devices such as FPGAs), a memory 930 for storing data, and one or more storage media 920 (e.g., one or more mass storage devices) for storing application programs 923 or data 922. The memory 930 and storage media 920 may be temporary or persistent storage. The program stored in the storage media 920 may include one or more modules, each module may include a series of instruction operations on the electronic device. Furthermore, the CPU 910 may be configured to communicate with the storage media 920 and execute the series of instruction operations in the storage media 920 on the electronic device 900. Electronic device 900 may also include one or more power supplies 960, one or more wired or wireless network interfaces 950, one or more input / output interfaces 940, and / or one or more operating systems 921, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.

[0139] The input / output interface 940 can be used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of the electronic device 900. In one example, the input / output interface 940 includes a network interface controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the input / output interface 940 may be a radio frequency (RF) module for wireless communication with the Internet.

[0140] Embodiments of this application also provide a computer-readable storage medium, which can be disposed in an electronic device to store at least one instruction or at least one program related to implementing a data center control method in the method embodiments. The at least one instruction or the at least one program is loaded and executed by the processor to implement the data center control method provided in the above method embodiments.

[0141] Optionally, in this embodiment, the storage medium may be located at at least one of the multiple network servers in a computer network. Optionally, in this embodiment, the storage medium may include, but is not limited to, various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0142] According to one aspect of this application, a computer program product or computer program is provided, comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the methods provided in the various alternative implementations described above.

[0143] It should be understood that this application is not limited to the processes and structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A computer room control method, characterized in that, include: The configuration information of the refrigeration equipment modules pre-integrated within the equipment container is obtained, and the configuration information is parsed to generate basic refrigeration capacity characterization information of the equipment container. Obtain rack deployment information of one or more racks deployed on-site in the modular data center, process the rack deployment information to generate cooling demand characterization information for the one or more racks; The basic cooling capacity characterization information and the cooling demand characterization information are matched and calculated to generate an initial cooling control strategy; The system acquires environmental status information collected by the monitoring and fire protection equipment module pre-integrated inside the equipment container, and modifies the initial cooling control strategy based on the environmental status information to generate a dynamically modified cooling control strategy. Based on the aforementioned dynamic correction cooling control strategy, the operating status of the cooling equipment module is controlled to control the cooling output of the prefabricated modular data center.

2. The computer room control method according to claim 1, characterized in that, The process of acquiring configuration information of pre-integrated refrigeration equipment modules within the equipment container and parsing the configuration information to generate basic refrigeration capacity characterization information for the equipment container includes: Extract the number and rated power of the air conditioning units and the number and rated power of the liquid-cooled CDU units included in the refrigeration equipment module from the configuration information; Based on the number and rated power of the air conditioning equipment and the number and rated power of the liquid-cooled CDU equipment, the maximum theoretical cooling capacity of the equipment container in air-cooled mode and liquid-cooled mode is calculated to generate the basic cooling capacity characterization information.

3. The computer room control method according to claim 1, characterized in that, The step of acquiring rack deployment information of one or more racks deployed on-site in the modular data center, and processing the rack deployment information to generate cooling demand characterization information for the one or more racks, includes: Extract the total number of deployed cabinets, the power rating of each cabinet, and the location coordinates of each cabinet within the data center from the cabinet deployment information. The total heat load requirement is calculated based on the total number of deployed cabinets and the power rating of each cabinet; Based on the location coordinates of each cabinet, the total heat load demand is decomposed into regional heat load demands corresponding to each air conditioning supply area or liquid cooling circuit partition in the refrigeration equipment module, so as to generate the refrigeration demand characterization information containing the total heat load demand and the regional heat load demand.

4. The computer room control method according to claim 3, characterized in that, The step of obtaining rack deployment information for one or more racks deployed on-site in the prefabricated modular data center includes: Based on the current total business load planned for the prefabricated modular data center and the preset standard load capacity of a single rack, calculate the target number of racks required to meet the current total business load. Based on the target number of server racks and the pre-set rack layout template of the prefabricated modular data center, the rack deployment information is generated.

5. The computer room control method according to claim 1, characterized in that, The step of matching and calculating the basic cooling capacity characterization information and the cooling demand characterization information to generate an initial cooling control strategy includes: The total heat load demand in the cooling demand characterization information is compared with the maximum theoretical cooling capacity in the basic cooling capacity characterization information to determine whether the activated cooling mode is air-cooled mode, liquid-cooled mode or hybrid mode. If it is determined that the air-cooled mode or the hybrid mode is to be enabled, the number of start-ups, operating power and air supply direction of each air conditioner in the refrigeration equipment module are calculated and allocated according to the regional heat load demand in the refrigeration demand characterization information, so as to generate the first part of the control instructions related to the air conditioner in the initial refrigeration control strategy. If it is determined that liquid cooling mode or hybrid mode is enabled, the number of liquid-cooled CDU devices to be started, the operating power and the corresponding secondary side loop flow rate of each liquid-cooled CDU device in the refrigeration equipment module are calculated and allocated according to the regional heat load demand in the refrigeration demand characterization information, so as to generate the second part of the control instructions related to the liquid-cooled equipment in the initial refrigeration control strategy.

6. The computer room control method according to claim 1, characterized in that, The step of acquiring environmental status information collected by the pre-integrated monitoring and fire protection equipment module inside the equipment container, and modifying the initial cooling control strategy based on the environmental status information to generate a dynamically modified cooling control strategy, includes: Extract the external ambient temperature of the computer room, the real-time temperature of different monitoring points inside the computer room, and the real-time power consumption data of the cabinet fed back by the power distribution equipment module from the environmental status information. Based on the comparison between the external ambient temperature of the computer room and the preset temperature threshold, the operating mode setting of the air conditioning equipment in the initial cooling control strategy is adjusted. Based on the deviation between the real-time temperature at different monitoring points inside the computer room and the corresponding area heat load demand in the cooling demand characterization information, the operating power of the air conditioning equipment in the corresponding area or the flow rate of the corresponding liquid cooling circuit in the initial cooling control strategy is adjusted proportionally-integrally-derivatively. Based on the difference between the real-time power consumption data of the cabinet and the corresponding cabinet power rating in the cabinet deployment information, the regional heat load demand in the cooling demand characterization information is dynamically updated, and the initial cooling control strategy is recalculated based on the updated regional heat load demand. The adjusted, proportional-integral-derivative (PI-DE) control and the recalculated control commands are fused together to generate the dynamically corrected refrigeration control strategy.

7. The computer room control method according to claim 1, characterized in that, The method of controlling the operating status of the cooling equipment module based on the dynamically corrected cooling control strategy to control the cooling output of the prefabricated modular data center includes: According to the dynamic correction refrigeration control strategy, control signals including operating mode, target power and air supply angle are sent to the air conditioning equipment in the refrigeration equipment module, and control signals including target power and target flow rate are sent to the liquid-cooled CDU equipment in the refrigeration equipment module. According to the dynamic correction cooling control strategy, the on / off state and power distribution of the power supply circuit associated with the cooling equipment module in the power distribution equipment module are adjusted, and the monitoring and fire protection equipment module is activated to continuously monitor the adjusted cooling output effect and the computer room environmental parameters, so as to complete the closed-loop control of the cooling output of the prefabricated modular data center.

8. The computer room control method according to claim 4, characterized in that, The process of generating rack deployment information based on the target number of racks and the pre-set rack layout template of the prefabricated modular data center includes: Available rack locations that meet the power capacity constraints and the cooling capacity constraints of the corresponding cooling area are selected from the rack layout template, and multiple candidate location subsets are constructed based on the target number of racks; For each candidate location subset, the expected new heat load generated by the candidate location subset in each cooling zone is determined based on the power rating of the cabinet to be deployed and the dynamic load coefficient representing the corresponding service load level. Based on the expected new heat load, the existing heat load of each cooling zone and the cooling capacity of each cooling zone, the cooling capacity utilization rate of each cooling zone after the cabinet is deployed is determined. The degree of dispersion of the cooling capacity utilization rate corresponding to the candidate location subset is determined based on the cooling capacity utilization rate corresponding to each cooling zone, and the airflow organization score is determined based on the influence of the candidate location subset on the continuity of the cold and hot aisle isolation. From the candidate location subset where the cooling capacity utilization rate of each cooling zone is lower than the preset safety threshold, the target location subset is determined based on the dispersion of the cooling capacity utilization rate and the airflow organization score, and the cabinet deployment information is generated based on the target location subset.

9. The computer room control method according to claim 7, characterized in that, The adjustment of the on / off state and power distribution of the power supply circuit associated with the refrigeration equipment module in the power distribution equipment module includes: Before disconnecting the power supply circuit associated with the target refrigeration equipment to be powered off, the current operating status of the target refrigeration equipment is obtained, and the power outage lockout for the power supply circuit is released when the current operating status belongs to a preset power outage allowed state. Before starting the target refrigeration device, the current real-time load power, rated capacity of the power supply circuit associated with the target refrigeration device, and the peak power of the target refrigeration device at startup are obtained. The estimated total load after startup is determined based on the current real-time load power and the peak power at startup. The startup lockout for the target refrigeration device is released when the estimated total load is not greater than the product of the rated capacity of the circuit and the preset safety factor. After the corresponding lockout is released, the on / off control of the power supply circuit or the start control of the target refrigeration equipment is executed, and closed-loop feedback data characterizing the cooling output effect is acquired at a monitoring frequency higher than the refrigeration control cycle. The closed-loop feedback data includes at least two of the following: air conditioner outlet temperature, liquid cooling circuit supply and return water temperature difference, and temperature change rate of each refrigeration zone. When a key parameter in the closed-loop feedback data deviates from the expected range to a preset deviation threshold, the cabinet deployment information is reacquired, or the dynamic correction cooling control strategy is recalculated based on the latest closed-loop feedback data.

10. A modular data center, characterized in that, The computer room control method applicable to any one of claims 1 to 9 includes: Supporting devices for equipment containers, one or more cabinets, and low-voltage cable trays; The equipment container is pre-integrated with refrigeration equipment modules, power distribution equipment modules, and monitoring and fire protection equipment modules. The one or more cabinets are arranged independently of the equipment container and are detachably connected to the refrigeration equipment module and power distribution equipment module inside the equipment container; The support device is rotatably mounted on the top or side of the equipment container; The support device has a retracted state and an extended state: in the retracted state, the support device is rotated and stored in the internal space of the equipment container or close to the outer contour; in the extended state, the support device is rotated out from the equipment container and located above the one or more cabinets, for supporting and fixing the low-voltage cable tray.

Images