Dynamic capacity configuration method and device based on physical information digital twinning
By simulating the thermodynamic processes of data centers, predicting energy consumption and carbon emissions, and optimizing the operational carbon footprint, this technology solves the problem of the inability to achieve precise capacity management in complex environments in existing technologies, and realizes efficient data center resource allocation and energy management.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-08
- Publication Date
- 2026-04-03
AI Technical Summary
Existing energy modeling models mainly focus on the energy consumption of computing devices. Through experimental data verification, they can capture energy consumption and the carbon footprint of data centers. However, since they only consider historical data from stable operating points, their ability to predict and extrapolate actual data center energy consumption patterns in complex and ever-changing environments is limited, and they cannot achieve accurate capacity management.
A dynamic capacity configuration method based on physical information digital twins is adopted to predict energy consumption and carbon emissions by simulating the thermodynamic processes of data centers, optimize the carbon footprint of operation, generate capacity budgets, and dynamically allocate resources.
It enables precise capacity management of data centers in complex and ever-changing environments, reduces energy waste, and improves forecasting accuracy and operational efficiency.
Smart Images
Figure CN121785750A_ABST
Abstract
Description
Technical Field
[0001] This specification relates to the field of data center energy consumption optimization technology, and in particular to a dynamic capacity configuration method and device based on physical information digital twins. Background Technology
[0002] Energy modeling is the foundation of carbon-aware data center management, enabling the capture of carbon consumption and the data center's carbon footprint to predict data center energy consumption for energy allocation.
[0003] In existing technologies, energy modeling mainly focuses on the energy consumption of computing devices, such as linear models and simplified empirical implementation models. Existing energy modeling models can capture energy consumption and the carbon footprint of data centers through experimental data validation. However, because they only consider historical data from stable operating points, their ability to predict and extrapolate actual data center energy consumption patterns is limited. Therefore, they cannot achieve accurate capacity management of data centers in complex and ever-changing environments.
[0004] Therefore, a new method for dynamic capacity configuration in data centers is needed. Summary of the Invention
[0005] This specification provides a dynamic capacity configuration method and apparatus based on physical information digital twins to address the following technical problems: In the prior art, energy modeling mainly focuses on the energy consumption of computing devices, such as linear models and simplified empirical implementation models. Existing energy modeling models, verified by experimental data, can capture energy consumption and the carbon footprint of data centers. However, because they only consider historical data from stable operating points, their ability to predict and extrapolate actual data center energy consumption patterns is limited. Therefore, they cannot achieve accurate capacity management of data centers in complex and ever-changing environments.
[0006] To solve the above-mentioned technical problems, the embodiments in this specification are implemented as follows:
[0007] This specification provides an embodiment of a dynamic capacity configuration method based on physical information digital twins, including:
[0008] A digital twin model of physical information is used to simulate the thermodynamic processes of the data center to be processed;
[0009] Based on the thermodynamic processes of the data center to be processed, predict the energy consumption and carbon emissions of the data center to be processed;
[0010] Based on the predicted energy consumption and carbon emissions of the data center to be processed, the operating carbon footprint of the data center to be processed is optimized to generate a capacity budget.
[0011] Based on the capacity budget, the resources of the data center to be processed are configured.
[0012] This specification also provides an embodiment of a dynamic capacity configuration device based on physical information digital twin, comprising:
[0013] The simulation module uses a digital twin model of physical information to simulate the thermodynamic processes of the data center to be processed;
[0014] The prediction module predicts the energy consumption and carbon emissions of the data center to be processed based on the thermodynamic processes of the data center to be processed.
[0015] The optimization module optimizes the operational carbon footprint of the data center to be processed based on the predicted results of its energy consumption and carbon emissions, and generates a capacity budget.
[0016] The configuration module configures the resources of the data center to be processed based on the capacity budget.
[0017] This specification provides a dynamic capacity configuration method and apparatus based on physical information digital twins. It employs a physical information digital twin model to simulate the thermodynamic processes of a data center to be processed; based on these thermodynamic processes, it predicts the energy consumption and carbon emissions of the data center; based on the predicted energy consumption and carbon emissions, it optimizes the operational carbon footprint of the data center to generate a capacity budget; and based on the capacity budget, it configures the resources of the data center. This method accurately simulates the data center, making its operation more efficient, reducing energy waste, and allowing for flexible adjustment of data center resources with high predictive accuracy. Attached Figure Description
[0018] To more clearly illustrate the technical solutions in the embodiments or prior art of this specification, the drawings used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0019] Figure 1 A flowchart illustrating a dynamic capacity configuration method based on physical information digital twins, provided for embodiments of this specification;
[0020] Figure 2 A schematic diagram of a digital twin model provided in the embodiments of this specification;
[0021] Figure 3 A framework diagram of a dynamic capacity configuration method based on physical information digital twin provided in the embodiments of this specification;
[0022] Figure 4 This is a schematic diagram of a dynamic capacity configuration device based on physical information digital twin, provided as an embodiment of this specification. Detailed Implementation
[0023] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification 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 specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this application.
[0024] Figure 1 This is a flowchart illustrating a dynamic capacity configuration method based on physical information digital twins, provided as an embodiment of this specification. From a programming perspective, the execution entity of the process can be a program mounted on an application server or application terminal. It is understood that this method can be executed by any device, equipment, platform, or cluster of devices with computing and processing capabilities. Figure 1 As shown, the dynamic capacity configuration method includes:
[0025] Step S101: Use a digital twin model of physical information to simulate the thermodynamic process of the data center to be processed.
[0026] In the embodiments described in this specification, the data center to be processed is the data center that requires dynamic capacity configuration.
[0027] To further understand the digital twin model provided in the embodiments of this specification, a detailed description of the digital twin model will be provided below. Figure 2 This is a schematic diagram of a digital twin model provided in an embodiment of this specification. Figure 2 As shown, the physical workflow of the digital twin model includes air circulation, cooling water circulation, and condensate circulation. The computational workflow of the digital twin model includes data hall air circulation thermodynamics based on air circulation, cooling water circulation thermodynamics based on cooling water circulation, and condensate circulation thermodynamics based on condensate circulation. The data hall air circulation thermodynamics is implemented based on the data hall air circulation thermodynamics model, which is based on the affinity law, thereby obtaining the device power consumption (P). CRAH The thermodynamics of the cooling water circulation and condensate circulation are realized based on the thermodynamic model of the chiller unit. This model is based on the law of affinity, and thus the energy consumption P of the cooling water pump is obtained. chwp Energy consumption P of chiller unitch P of the condenser chp and the P of the cooling unit ct .
[0028] In the embodiments described in this specification, in order to simulate the thermodynamic processes of a data center based on a digital twin model, system modeling is required first to capture the energy consumption of computing devices. Specifically, this includes workload modeling, aggregate computing demand modeling, and server modeling. The system modeling will be described in detail below.
[0029] The workload model primarily considers virtual machine (VM) workloads. At the fine-grained time slot k, N VM [k] Requests entry into the system. Each workload specifies its CPU core requirements and deadline.
[0030] The capacity planner, during advanced capacity configuration, is responsible for determining the total amount of computing resources required at each time point. Therefore, models using aggregated computing requirements can determine how to allocate computing resources based on the total resource requirements of all upcoming and currently running tasks. In the k-th time slot, it is defined as:
[0031]
[0032] Where N VM [k] is the number of incoming tasks in the k-th fine-grained time slot. It is the task submission slot index, l i It is the task's runtime, d i It refers to the resource requirements of the task.
[0033] When modeling the servers, we assume a data center with N homogeneous servers, each server having a total of c CPU cores. total The i-th server tracks the following three variables in each fine-grained time slot k: the number of cores used. Utilization Power consumption P i When a new task is assigned to a server or a task running on a server completes, the scheduler updates these three variables accordingly for each server. The power consumption of the i-th server is estimated using a linear power model:
[0034] P i [k]=P static +(P full -P static )·u i [k]
[0035] Where P static It is the server's static power, P fullThis is the power consumption when the server is fully utilized. The total power consumption of an IT system can be expressed as:
[0036] In the embodiments described in this specification, the digital twin model includes a data hall air circulation thermodynamic model, a chiller unit thermodynamic model, and an equipment power consumption model;
[0037] in,
[0038] The data hall air circulation thermodynamic model is a discrete-time model. The data hall air circulation thermodynamic model is based on the heat load of the data center to be processed and the operating settings of each CRAH unit, and is used to predict the average temperature of the next time slot.
[0039] The thermodynamic model of the chiller unit uses the NTU-efficiency method to simulate the heat transfer process. The thermodynamic model of the chiller unit uses a uniform load distribution mechanism to distribute the total cooling load to each chiller unit. The total cooling load is equal to the total heat load of the data center.
[0040] The device power consumption model is used to determine the power consumption of each facility in the data center to be processed.
[0041] In the embodiments described in this specification, the CRAH unit refers to the computer room air handling unit.
[0042] In the embodiments of this specification, the data hall air circulation thermodynamic model simulates the dynamic process of heat exchange in the data hall by setting the supply air temperature, mass flow rate and total heat load of the CRAH unit in the data center to be processed.
[0043] The thermodynamic model of air circulation in the data hall is as follows: Among them, T z [k+1] represents the predicted average temperature of the (k+1)th time slot, N CRAH N represents the number of active CRAC units in the data center to be processed, and i represents the number of active CRAC units i in the data center to be processed. pa This indicates the specific heat capacity of water. This represents the mass flow rate of the i-th CRAC unit in the k-th time slot. T represents the supply air temperature of the i-th CRAC unit in the k-th time slot. z [k] represents the predicted average temperature of the k-th time slot, and Q[k] represents the predicted total heat load of the k-th time slot;
[0044] The cooling load of the i-th chiller unit in the data center to be processed, as determined by the thermodynamic model of the chiller unit, is... in, Let N represent the cooling load of the i-th chiller unit in the k-th time slot, Q[k] represent the total heat load in the k-th time slot, and N represent the total heat load in the k-th time slot. ch This indicates the number of chiller units in the data center.
[0045] The energy consumption of the k-th time slot of the data center to be processed, as determined by the device energy consumption model, is: P CRAH [k]=∑ i＝1 f C ′ RAH (m sup[ki] ),P chwp [k]=f chwp (m chwp [k]), where P CRAH [k] represents the energy consumption of the k-th time slot of the data center to be processed. Let m represent the power consumption function of the i-th CRAH. sup [k]) represents the supply air volume in the k-th time slot, P chwp [k] represents the energy consumption of the cooling water pump in the k-th time slot, f chwp This represents the power consumption function for calculating the cooling water pump, m chwp [k] represents the cooling water pump flow rate in the k-th time slot.
[0046] In the embodiments of this specification, a data hall air circulation thermodynamic model is used to characterize the heat transfer process inside the data center to be processed. In the data hall air circulation thermodynamic model, the temperature in the data center is uniformly distributed; therefore, the average room temperature and airflow are sufficient. Let N... CRAH This indicates the number of active CRAC units within the data center. and Let represent the supply air temperature and mass flow rate of the i-th CRAC unit at time t, respectively. Furthermore, UPS electrical efficiency is considered, and the total heat load of the data center is modeled as... Where α is the UPS electrical efficiency. α is continuously updated as a moving average of the online data. Thermodynamics can be modeled using the following ordinary differential equation (ODE): in It is the specific heat capacity of air, T z (t) is the temperature of the region at time t, thus obtaining the data hall air circulation thermodynamic model provided in the embodiments of this specification.
[0047] In a specific embodiment, based on the data hall air circulation thermodynamic model provided in the embodiments of this specification, the heat load of the data center to be processed and the operation of the CRAH unit can be set, and the average room temperature of the next time slot can be predicted.
[0048] In the embodiments described in this specification, the chiller unit consists of a cooling water circuit and a condenser water circuit. Both loops have a supply-side half-loop and a demand-side half-loop. The cooling water demand-side half-loop connects to the cooling coils and cooler of each CRAH unit.
[0049] The cooling water demand-side half-loop connects the cooling coils of the CRAC unit and the chiller unit, therefore it is necessary to determine the heat load of each cooling coil and its chilled water flow rate. In the embodiments of this specification, a uniform load setting Q[k] is used to allocate the historical operating data observed from the k-th time slot to the cooling coils. For the cooling coil of the i-th CRAC unit, its cooling load is... in, Let Q[k] be the cooling load of the cooling coil of the i-th CRAC unit, and N be the set load. CRAH This indicates the number of active CRAC units within the data center. Further, the cooling coils transfer the allocated heat load to the chilled water. To calculate the required chilled water mass flow rate to meet the cooling load, the embodiments in this specification employ the NTU-efficiency method to simulate the heat transfer process. Specifically, a cooling water flow rate of [missing information] is selected. in This is the specific heat capacity of water. Under this setting, the heat transfer efficiency between air and chilled water is 1. Therefore, the total chilled water mass flow rate is
[0050] The cooling water supply side semi-loop includes a chiller unit to supply cooling water to the CRAC unit, and a mechanical pump circulates the cooling water in the cooling water loop. In the embodiments of this specification, N is provided. ch There are 11 chiller units. The total cooling load of all chiller units equals the total heat load of the data center. A uniform load distribution mechanism is used to allocate the total cooling load to each chiller unit. The cooling load of the i-th chiller unit is... The partial load ratio (PLR) of the i-th chiller unit is defined as follows: in This represents the cooling capacity of the i-th chiller unit. The pump's mass flow rate is the total cooling water mass flow rate (m) of the demand-side half-loop. chw [k].
[0051] The condenser water loop connects the cooler and cooling tower via a fluid loop. Its structure is similar to a chilled water loop. The heat load of the cooler is transferred to the condenser water through its condenser, and then discharged into the environment through the cooling tower. The condenser water pump and cooling tower typically operate at their rated speeds, with the cooling tower and condenser water pump set to operate at their rated settings.
[0052] It should be noted that the digital twin model constructed in the embodiments of this specification can achieve real-time online data updates.
[0053] Step S103: Based on the thermodynamic processes of the data center to be processed, predict the energy consumption and carbon emissions of the data center to be processed.
[0054] In the embodiments of this specification, predicting the energy consumption and carbon emissions of the data center to be processed based on its thermodynamic processes specifically includes:
[0055] Based on the thermodynamic processes of the data center to be processed, a predictive model is used to predict the energy consumption of the data center to be processed, and a data center carbon footprint model is used to predict the carbon emissions of the data center to be processed.
[0056] In the embodiments described in this specification, the data center carbon footprint model is: c[k]=ρ[k]·[P dc (u[k];e[k],θ)·r[k]] + ,in[] + Let max(0,·) represent the grid carbon intensity in the k-th period, θ be the set of all learnable parameters in the digital twin, u[k] represent the capacity supply plan in the k-th time slot, e[k] represent the set of control inputs and thermal load conditions in the k-th time slot, and r[k] represent the renewable supply in the k-th time slot.
[0057] P dc (u[k]; e[k], θ) represents the total power of the data center to be processed, P dc (u[k];e[k],θ)=P IT (u[k])+P elg (Q(u[k]); e[k],θ), P IT (u[k]) represents the power consumption of the IT system in the k-th time slot; Q(u[k]) represents the quantified value of the cooling or energy consumption requirement derived from the IT system's capacity configuration u[k]. This value reflects how much cooling resources are needed to handle the heat generated by the IT equipment. clg (Q(u[k]);e[k],θ) represents the cooling system power consumption derived from the IT equipment cooling demand Q(u[k]) under the influence of environmental and operating conditions e[k] and the model's learnable parameter θ.
[0058] The set of control inputs and thermal load conditions, e = [T] sup ,m sup ,T chw ,T cw ,T o ], T supRepresents the supply temperature, m sup T represents the supply air volume. chw T represents the cooling water supply temperature. cw T represents the temperature of the condensate. o This represents the outdoor wet-bulb temperature.
[0059] In the embodiments described in this specification, the prediction model includes an IT workload predictor, a workload uptime predictor, and a solar energy predictor.
[0060] The IT workload predictor uses hot-coded time features as input to a Poisson arrival model to obtain the number of workloads arriving within a preset time period; it then uses the number of workloads arriving within the preset time period and the time features as input to a resource model to predict the resource requests for each workload, where the resource model is an LSTM model; finally, it uses the predicted resource requests for each workload and the time features as input to a duration model to predict the runtime of each workload, where the duration model is also an LSTM model.
[0061] The workload runtime predictor takes the current resource requirements as input and outputs a probability vector. Based on the upper and lower bounds represented by the index of the probability vector sampling, the actual running time is predicted, and the probability vector... This is prior knowledge of the running time of the i-th task. It is a d-dimensional real space;
[0062] The solar energy predictor uses weather variables and time characteristics as covariates and outputs solar energy prediction results based on the DeepAR time series prediction model.
[0063] In the embodiments described in this specification, the time features are hours and days of the week. The expression for the Poisson arrival model...
[0064]
[0065] In the embodiments described in this specification, weather variables include: solar radiance, outdoor dry-bulb temperature, and wind speed.
[0066] Step S105: Based on the predicted energy consumption and carbon emissions of the data center to be processed, optimize the operating carbon footprint of the data center to be processed and generate a capacity budget.
[0067] In the embodiments of this specification, the optimization of the operational carbon footprint of the data center to be processed based on the predicted energy consumption and carbon emissions of the data center to be processed, and the generation of a capacity budget, specifically includes:
[0068] Based on the predicted energy consumption and carbon emissions of the data center to be processed, minimizing the carbon footprint of the data center to be processed is taken as the optimization objective. The task running time is re-estimated using the constraint of the conservation of computing resource supply and the lower bound of computing resources as constraints. The computing resources are optimized based on MPC, and a capacity budget is generated according to the aggregated computing demand model.
[0069] In the embodiments described in this specification, the optimization objective of the MPC is:
[0070]
[0071]
[0072]
[0073] |u[k+1]-u[k]|≤σ.
[0074] Based on the optimization objective of the MPC, the first optimized action u * [τ] serves as the capacity budget;
[0075] in,
[0076] u[τ] and u[H] represent capacity configuration decision variables at different round numbers (τ to H);
[0077] ρ[k] represents the grid carbon intensity in the k-th time slot;
[0078] P dc (u[k];e[k],θ) represents the total energy consumption of the data center to be processed;
[0079] This represents the predicted renewable energy supply for the k-th time slot;
[0080] s[τ] represents the cumulative supply deficit up to time τ, which is the sum of the supply shortages in the past period;
[0081] u[k] represents the capacity configuration decision in the k-th time slot;
[0082] This represents the predicted computational requirement in the k-th time slot;
[0083] ψ represents the transfer / translation ratio.
[0084] C = Nc total Represents total computing resources.
[0085] σ represents the maximum permissible amount of configuration change between two consecutive time slots.
[0086] In the embodiments described in this specification, the re-estimation of task runtime includes:
[0087] If the task execution time is underestimated, the task execution time is updated using the posterior probability, where the prior probability is... l i is the task runtime, m represents the number of fine-grained units, and Pr represents the probability distribution used to update the task runtime;
[0088] If the task execution time is overestimated, the overestimated value of the task execution time is deducted to update the task execution time;
[0089] The configuration of computing resources is determined based on the aggregate computing demand model, which is as follows:
[0090] Where N VM [k] is the number of incoming tasks in the k-th fine-grained time slot. It is the task submission slot index, l i It is the task's runtime, d i It refers to the resource requirements of the task.
[0091] Step S107: Configure the resources of the data center to be processed based on the capacity budget.
[0092] For data centers awaiting processing, resources are configured based on capacity budgets. Simultaneously, during the configuration process, a capacity planner adaptively monitors and dynamically adjusts the capacity budget, thereby achieving the goal of dynamically adjusting resource allocation for data centers awaiting processing and ultimately minimizing carbon footprint.
[0093] The dynamic capacity configuration method based on physical information digital twins provided in this specification employs a physical information digital twin model to simulate the thermodynamic processes of a data center to be processed; based on the thermodynamic processes of the data center to be processed, the energy consumption and carbon emissions of the data center to be processed are predicted; based on the predicted energy consumption and carbon emissions of the data center to be processed, the operating carbon footprint of the data center to be processed is optimized to generate a capacity budget; based on the capacity budget, the resources of the data center to be processed are configured. This method can accurately simulate the data center, making the operation of the data center more efficient, reducing energy waste, and flexibly adjusting data center resources, with high predictive accuracy.
[0094] The above describes in detail a dynamic capacity configuration method based on physical information digital twins. Correspondingly, this specification also provides a dynamic capacity configuration device based on physical information digital twins, such as... Figure 4 As shown. Figure 4 This specification provides a schematic diagram of a dynamic capacity configuration device based on physical information digital twins, which includes:
[0095] Simulation module 401 uses a digital twin model of physical information to simulate the thermodynamic processes of the data center to be processed;
[0096] The prediction module 403 predicts the energy consumption and carbon emissions of the data center to be processed based on the thermodynamic processes of the data center to be processed.
[0097] The optimization module 405 optimizes the operating carbon footprint of the data center to be processed based on the predicted results of energy consumption and carbon emissions, and generates a capacity budget.
[0098] Configuration module 407 configures the resources of the data center to be processed based on the capacity budget.
[0099] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.
[0100] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments for apparatus, electronic devices, and non-volatile computer storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.
[0101] The apparatus, electronic device, and non-volatile computer storage medium and method provided in the embodiments of this specification are corresponding. Therefore, the apparatus, electronic device, and non-volatile computer storage medium also have similar beneficial technical effects as the corresponding method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the corresponding apparatus, electronic device, and non-volatile computer storage medium will not be repeated here.
[0102] In the 1990s, improvements to a technology could be clearly distinguished as either hardware improvements (e.g., improvements to the circuit structure of diodes, transistors, switches, etc.) or software improvements (improvements to the methodology). However, with technological advancements, many methodological improvements today can be considered direct improvements to the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved methodology into the hardware circuit. Therefore, it cannot be said that a methodological improvement cannot be implemented using hardware physical modules. For example, a Programmable Logic Device (PLD) (such as a Field Programmable Gate Array (FPGA)) is such an integrated circuit whose logic function is determined by the user programming the device. Designers can program and "integrate" a digital system onto a PLD themselves, without needing chip manufacturers to design and manufacture dedicated integrated circuit chips. Furthermore, nowadays, instead of manually manufacturing integrated circuit chips, this programming is mostly implemented using "logic compiler" software. Similar to the software compiler used in program development, the original code before compilation must be written in a specific programming language, called a Hardware Description Language (HDL). There are many HDLs, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, and RHDL (Ruby Hardware Description Language). Currently, the most commonly used are VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should understand that by simply performing some logic programming on the method flow using one of these hardware description languages and programming it into an integrated circuit, the hardware circuit implementing the logical method flow can be easily obtained.
[0103] The controller can be implemented in any suitable manner. For example, it can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicon Labs C8051F320. A memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art will also recognize that, in addition to implementing the controller in purely computer-readable program code form, the same functionality can be achieved by logically programming the method steps to make the controller take the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the means included therein for implementing various functions can also be considered as structures within the hardware component. Alternatively, the means for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.
[0104] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.
[0105] For ease of description, the above apparatus is described by dividing it into various functional units. Of course, when implementing one or more embodiments of this specification, the functions of each unit can be implemented in one or more software and / or hardware.
[0106] Those skilled in the art will understand that the embodiments of this specification can be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the embodiments of this specification can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0107] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0108] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0109] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0110] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0111] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.
[0112] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.
[0113] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0114] This specification can be described in the general context of computer-executable instructions that are executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This specification can also be practiced in distributed computing environments, where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside on local and remote computer storage media, including storage devices.
[0115] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.
[0116] The above description is merely an embodiment of this specification and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of this application should be included within the scope of the claims of this application.
Claims
1. A dynamic capacity configuration method based on physical information digital twins, characterized in that, The dynamic capacity configuration method includes: A digital twin model of physical information is used to simulate the thermodynamic processes of the data center to be processed; Based on the thermodynamic processes of the data center to be processed, predict the energy consumption and carbon emissions of the data center to be processed; Based on the predicted energy consumption and carbon emissions of the data center to be processed, the operating carbon footprint of the data center to be processed is optimized to generate a capacity budget. Based on the capacity budget, the resources of the data center to be processed are configured.
2. The dynamic capacity allocation method as described in claim 1, characterized in that, The digital twin model includes a data hall air circulation thermodynamic model, a chiller unit thermodynamic model, and an equipment power consumption model; in, The data hall air circulation thermodynamic model is a discrete-time model. The data hall air circulation thermodynamic model is based on the heat load of the data center to be processed and the operating settings of each CRAH unit, and is used to predict the average temperature of the next time slot. The thermodynamic model of the chiller unit uses the NTU-efficiency method to simulate the heat transfer process. The thermodynamic model of the chiller unit uses a uniform load distribution mechanism to distribute the total cooling load to each chiller unit. The total cooling load is equal to the total heat load of the data center. The device power consumption model is used to determine the power consumption of each facility in the data center to be processed.
3. The dynamic capacity allocation method as described in claim 1, characterized in that, The data hall air circulation thermodynamic model simulates the dynamic process of heat exchange within the data hall by setting the supply air temperature, mass flow rate, and total heat load of the CRAH units in the data center to be processed. The thermodynamic model of air circulation in the data hall is as follows: Among them, T z [k+1] represents the predicted average temperature of the (k+1)th time slot, N CRAH N represents the number of active CRAC units in the data center to be processed, and i represents the number of active CRAC units i in the data center to be processed. pa This indicates the specific heat capacity of water. T represents the mass flow rate of the i-th CRAC unit in the k-th time slot. i sup [k] represents the supply air temperature of the i-th CRAC unit in the k-th time slot, T z [k] represents the predicted average temperature of the k-th time slot, and Q[k] represents the predicted total heat load of the k-th time slot; The cooling load of the i-th chiller unit in the data center to be processed, as determined by the thermodynamic model of the chiller unit, is... in, Let N represent the cooling load of the i-th chiller unit in the k-th time slot, Q[k] represent the total heat load in the k-th time slot, and N represent the total heat load in the k-th time slot. ch This indicates the number of chiller units in the data center; The energy consumption of the k-th time slot of the data center to be processed, as determined by the equipment energy consumption model, is: P chwp [k]=f chwp (m chwp [k]), where P CRAH [k] represents the energy consumption of the k-th time slot of the data center to be processed. Let m represent the power consumption function of the i-th CRAH. sup [k]) represents the supply air volume in the k-th time slot, P chwp [k] represents the energy consumption of the cooling water pump in the k-th time slot, f chwp This represents the power consumption function for calculating the cooling water pump, m chwp [k] represents the cooling water pump flow rate in the k-th time slot.
4. The dynamic capacity allocation method as described in claim 1, characterized in that, The prediction of energy consumption and carbon emissions of the data center to be processed based on its thermodynamic processes specifically includes: Based on the thermodynamic processes of the data center to be processed, a predictive model is used to predict the energy consumption of the data center to be processed, and a data center carbon footprint model is used to predict the carbon emissions of the data center to be processed.
5. The dynamic capacity configuration method as described in claim 4, characterized in that, The data center carbon footprint model is: c[k]=ρ[k]·[P] dc (u[k];e[k],θ)·r[k]] + ,in[] + Let max(0,·) represent the grid carbon intensity in the k-th period, θ be the set of all learnable parameters in the digital twin, u[k] represent the capacity supply plan in the k-th time slot, e[k] represent the set of control inputs and thermal load conditions in the k-th time slot, and r[k] represent the renewable supply in the k-th time slot. P dc (u[k]; e[k], θ) represents the total power of the data center to be processed, P dc (u[k];e[k],θ)=P IT (u[k])+P elg (Q(u[k]); e[k],θ), P IT (u[k]) represents the power consumption of the IT system in the k-th time slot; Q(u[k]) represents the quantified value of the cooling or energy consumption requirement derived from the IT system's capacity configuration u[k]. This value reflects how much cooling resources are needed to handle the heat generated by the IT equipment. clg (Q(u[k]);e[k],θ) represents the cooling system power consumption derived from the IT equipment cooling demand Q(u[k]) under the influence of environmental and operating conditions e[k] and the model's learnable parameter θ; The set of control inputs and thermal load conditions, e = [T] sup ,m sup ,T chw ,T cw ,T o ], T sup Represents the supply temperature, m sup T represents the supply air volume. chw T represents the cooling water supply temperature. cw T represents the temperature of the condensate. o This represents the outdoor wet-bulb temperature.
6. The dynamic capacity configuration method as described in claim 4, characterized in that, The prediction model includes an IT workload predictor, a workload uptime predictor, and a solar energy predictor. The IT workload predictor uses hot-coded time features as input to a Poisson arrival model to obtain the number of workloads arriving within a preset time period; it then uses the number of workloads arriving within the preset time period and the time features as input to a resource model to predict the resource requests for each workload, where the resource model is an LSTM model; finally, it uses the predicted resource requests for each workload and the time features as input to a duration model to predict the runtime of each workload, where the duration model is also an LSTM model. The workload runtime predictor takes the current resource requirements as input and outputs a probability vector. Based on the upper and lower bounds represented by the index of the probability vector sampling, the actual running time is predicted, and the probability vector... This is prior knowledge of the running time of the i-th task. It is a d-dimensional real space; The solar energy predictor uses weather variables and time characteristics as covariates and outputs solar energy prediction results based on the DeepAR time series prediction model.
7. The dynamic capacity allocation method as described in claim 1, characterized in that, Based on the predicted energy consumption and carbon emissions of the data center to be processed, the operational carbon footprint of the data center to be processed is optimized to generate a capacity budget, specifically including: Based on the predicted energy consumption and carbon emissions of the data center to be processed, minimizing the carbon footprint of the data center to be processed is taken as the optimization objective. The task running time is re-estimated using the constraint of the conservation of computing resource supply and the lower bound of computing resources as constraints. The computing resources are optimized based on MPC, and a capacity budget is generated according to the aggregated computing demand model.
8. The dynamic capacity configuration method as described in claim 7, characterized in that, The optimization objective of the MPC is: |u[k+1]-u[k]|≤σ. Based on the optimization objective of the MPC, the first optimized action u * [τ] serves as the capacity budget; in, u[τ] and u[H] represent capacity configuration decision variables at different round numbers (τ to H); ρ[k] represents the grid carbon intensity in the k-th time slot; P dc (u[k];e[k],θ) represents the total energy consumption of the data center to be processed; This represents the predicted renewable energy supply for the k-th time slot; s[τ] represents the cumulative supply deficit up to time τ, which is the sum of the supply shortages in the past period; u[k] represents the capacity configuration decision in the k-th time slot; This represents the predicted computational requirement in the k-th time slot; ψ represents the transfer / translation ratio. C = Nc total Represents total computing resources. σ represents the maximum permissible amount of configuration change between two consecutive time slots.
9. The dynamic capacity allocation method as described in claim 7, characterized in that, The re-estimation of task runtime includes: If the task execution time is underestimated, the task execution time is updated using the posterior probability, where the prior probability is... l i is the task runtime, m represents the number of fine-grained units, and Pr represents the probability distribution used to update the task runtime; If the task execution time is overestimated, the overestimated value of the task execution time is deducted to update the task execution time; The configuration of computing resources is determined based on the aggregate computing demand model, which is as follows: Where N VM [k] is the number of incoming tasks in the k-th fine-grained time slot. It is the task submission slot index, l i It is the task's runtime, d i It refers to the resource requirements of the task.
10. A dynamic capacity configuration device based on physical information digital twin, characterized in that, The dynamic capacity configuration device includes: The simulation module uses a digital twin model of physical information to simulate the thermodynamic processes of the data center to be processed; The prediction module predicts the energy consumption and carbon emissions of the data center to be processed based on the thermodynamic processes of the data center to be processed. The optimization module optimizes the operational carbon footprint of the data center to be processed based on the predicted results of its energy consumption and carbon emissions, and generates a capacity budget. The configuration module configures the resources of the data center to be processed based on the capacity budget.