Space-time scene information load energy consumption prediction method and related device

Abstract

The invention belongs to a prediction method, and provides a space-time scene information load energy consumption prediction method and a related device in order to solve the technical problems that information load energy consumption and nonlinear relations among numerous complex factors are difficult to capture and dynamic characteristics of the information load energy consumption along with time and space change exist in a traditional energy consumption prediction method. Timestamps in calculation task feature data, environment data, historical energy consumption data, spatial data and time data are obtained to form collection data, principal component extraction is performed after the collection data are processed, and principal components are obtained. The current energy consumption is obtained by means of an energy consumption model in a multivariate nonlinear regression form, the temperature rise caused by the energy consumption of the data center is obtained by means of a thermodynamic calculation model, and an energy consumption prediction result is obtained by taking the principal component, the current energy consumption and the temperature rise caused by the energy consumption of the data center as input of an energy consumption prediction model. And the precision and efficiency of data center energy efficiency management are obviously improved.

Description

Technical Field

[0001] The present application relates to a prediction method, specifically a method for predicting energy consumption of spatiotemporal scene information load and related devices. Background Art

[0002] As the infrastructure of the information age, data centers shoulder the crucial tasks of data storage, processing, and transmission. Consequently, their energy consumption is becoming increasingly prominent. Data center energy consumption is not only directly linked to operating costs but also a significant source of global carbon emissions. Therefore, efficiently and accurately predicting and managing the energy consumption of information loads has become a major challenge for the industry.

[0003] Traditional energy consumption prediction methods, such as rule-based empirical models or simple linear regression, often fail to fully capture the nonlinear relationship between information load energy consumption and numerous complex factors, as well as its dynamic characteristics that change over time and space. Summary of the Invention

[0004] This application addresses the technical problems of traditional energy consumption prediction methods, which have difficulty in capturing the nonlinear relationship between information load energy consumption and numerous complex factors, as well as the dynamic characteristics of information load energy consumption changing with time and space, and provides a spatiotemporal scenario information load energy consumption prediction method and related devices.

[0005] In order to achieve the above objectives, this application adopts the following technical solutions: In the first aspect, the present application proposes a method for predicting energy consumption of spatiotemporal scene information load, comprising: Obtain computing task feature data, environmental data, historical energy consumption data, spatial data, and time stamps from time data to form collected data; Performing format unification and coding processing on the collected data to obtain processed data; performing standardization on the processed data to obtain standardized data; Performing principal component extraction on the standardized data to obtain principal components; According to the principal components, the current energy consumption is obtained by using an energy consumption model in the form of multiple nonlinear regression; Based on environmental data, computing task characteristic data, and current energy consumption, the temperature rise of the data center caused by energy consumption is obtained using a thermodynamic calculation model. The main components, current energy consumption, and temperature rise of the data center caused by energy consumption are input into the energy consumption prediction model to obtain the energy consumption prediction results.

[0006] Furthermore, the method for formatting and encoding the collected data includes: Classify the timestamps in time data and convert them into numerical format; Carry out coding and standardization processing on spatial data; Perform data distribution consistency processing on environmental data; Process missing values and outliers in computing task feature data.

[0007] Furthermore, the method for standardizing the processed data includes: The processed data were transformed into a standard normal distribution.

[0008] Furthermore, the method for extracting principal components from the standardized data includes: The covariance matrix is obtained by the following formula :

[0009] in, is the sample size, For the Standardized data, is the mean vector of the standardized data; The covariance matrix is expressed as follows Perform eigenvalue decomposition:

[0010] in, is the eigenvector matrix, is a diagonal matrix; The elements on the diagonal of the diagonal matrix are used as eigenvalues, and the eigenvectors corresponding to the eigenvalues with the largest preset number are selected to form the principal component matrix ; Project the normalized data onto the principal component matrix The principal components are obtained from the eigenvectors of .

[0011] Furthermore, the expression of the energy consumption model in the form of multiple nonlinear regression includes:

[0012] in, is energy consumption, As the basic energy consumption level, are the coefficients of the principal components, is the principal component variable, is the coefficient of the spatiotemporal variable, is the spatiotemporal scene variable, is the coefficient of the interaction term, is the error term.

[0013] Furthermore, the expression of the thermodynamic calculation model includes:

[0014] in, The temperature rise caused by energy consumption in the data center, is the net heat change, is the specific heat capacity of air, For the quality of air.

[0015] Furthermore, the energy consumption prediction model adopts a random forest regression model.

[0016] In a second aspect, the present application proposes a spatiotemporal scene information load energy consumption prediction system, comprising: The data acquisition module is used to obtain computing task feature data, environmental data, historical energy consumption data, spatial data, and time stamps in time data to form collected data; A data processing module, configured to perform format unification and encoding processing on the collected data to obtain processed data; A standardization module, configured to perform standardization on the processed data to obtain standardized data; A principal component extraction module is used to extract principal components from the standardized data to obtain principal components; An energy consumption calculation module, configured to obtain current energy consumption based on the principal components using an energy consumption model in the form of multiple nonlinear regression; The temperature rise calculation module is used to calculate the temperature rise caused by energy consumption in the data center based on environmental data, computing task characteristic data, and current energy consumption using a thermodynamic calculation model. The prediction module is used to input the main components, current energy consumption, and temperature rise of the data center caused by energy consumption into the energy consumption prediction model to obtain energy consumption prediction results.

[0017] In the third aspect, the present application proposes an electronic device comprising: a memory, one or more processors; the memory is coupled to the processor; wherein computer program code is stored in the memory, and the computer program code includes computer instructions, and when the computer instructions are executed by the processor, the electronic device executes the steps of the above-mentioned spatiotemporal scene information load energy consumption prediction method.

[0018] In a fourth aspect, the present application proposes a computer-readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned spatiotemporal scene information load energy consumption prediction method are implemented.

[0019] Compared with the prior art, this application has the following beneficial effects: The present application proposes a method for predicting energy consumption of spatiotemporal scene information load, which obtains computing task characteristic data, environmental data, historical energy consumption data, timestamps in spatial data and time data to form collected data, processes the collected data and extracts principal components to obtain principal components. With the help of an energy consumption model in the form of multivariate nonlinear regression, the current energy consumption is obtained, and with the help of a thermodynamic calculation model, the temperature rise caused by energy consumption in the data center is obtained. The principal components, current energy consumption, and the temperature rise caused by energy consumption in the data center are used as inputs of the energy consumption prediction model to obtain energy consumption prediction results. The present application analyzes historical energy consumption data and environmental conditions, combines the influence of principal component variables and spatiotemporal variables, and uses energy consumption prediction models for training and optimization, thereby achieving accurate prediction of real-time energy consumption. It can accurately predict the energy consumption of different computing tasks, thereby optimizing energy utilization and equipment operation strategies, significantly improving the accuracy and efficiency of data center energy efficiency management, and effectively improving the energy efficiency and operational stability of data centers.

[0020] This application also proposes a spatiotemporal scene information load energy consumption prediction system, an electronic device and a computer storage medium, which have all the advantages of the above-mentioned spatiotemporal scene information load energy consumption prediction method. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0022] Figure 1 A flow chart of the method for predicting energy consumption of spatiotemporal scene information load in this application; Figure 2 This is a schematic diagram of the principle of an embodiment of the spatiotemporal scene information load energy consumption prediction method of the present application; Figure 3 A schematic diagram of the spatiotemporal scenario information load energy consumption prediction system of this application; DETAILED DESCRIPTION To make the objectives, technical solutions, and advantages of the embodiments of the present application more clear, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Generally, the components of the embodiments of the present application described and shown in the drawings herein can be arranged and designed in various different configurations.

[0023] Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the present application for protection, but merely represents selected embodiments of the present application. All other embodiments obtained by persons of ordinary skill in the art based on the embodiments in the present application without creative work are within the scope of protection of the present application.

[0024] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings.

[0025] In the description of the embodiments of the present application, it should be noted that if the terms "upper", "lower", "horizontal", "inner", etc. appear, the orientation or position relationship indicated is based on the orientation or position relationship shown in the accompanying drawings, or the orientation or position relationship in which the product of the invention is usually placed when in use. This is only for the convenience of describing the present application and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation. Therefore, it should not be understood as a limitation on the present application. In addition, the terms "first", "second", etc. are only used to distinguish the description and should not be understood as indicating or implying relative importance.

[0026] In addition, if the term "horizontal" appears, it does not mean that the component must be absolutely horizontal, but can be slightly tilted. For example, "horizontal" only means that its direction is more horizontal than "vertical", and does not mean that the structure must be completely horizontal, but can be slightly tilted.

[0027] In the description of the embodiments of this application, it should also be noted that, unless otherwise expressly specified or limited, the terms "disposed," "installed," "connected," and "connected" should be understood in a broad sense. For example, they can refer to fixed connections, detachable connections, or integral connections; they can refer to mechanical connections or electrical connections; they can refer to direct connections or indirect connections through an intermediate medium; and they can refer to internal connections between two components. Those skilled in the art can understand the specific meanings of the above terms in this application based on specific circumstances.

[0028] In today's booming digital economy, data centers, as core infrastructure supporting the flow of massive amounts of information and intelligent decision-making, face a critical bottleneck in their energy consumption, hindering the sustainable development of the industry. As the "energy black hole" of digital infrastructure, data centers not only account for over 30% of enterprise operating costs, but their electricity consumption also contributes a significant 2% of global carbon emissions, a figure that continues to grow. Faced with such a severe energy consumption challenge, building accurate prediction models and optimizing dynamic load management have become pressing technical challenges for the industry.

[0029] Compared to traditional prediction paradigms, such as empirical models that rely on artificial rules or simple linear regression algorithms, existing methods have significant flaws when dealing with data center energy consumption systems. First, complex nonlinear coupling exists between information load and multidimensional factors such as cooling systems, IT equipment, and power supply modules, making it difficult for traditional linear models to capture these high-order correlations. Second, energy consumption patterns exhibit multiple periodicities in the temporal dimension, such as diurnal fluctuations and seasonal cycles, while in the spatial dimension, they are subject to physical constraints such as rack layout and cooling paths. This dynamic nature of spatiotemporal coupling makes static prediction models difficult to adapt to real-world scenarios. This lack of predictive capability directly restricts the optimization space for the entire chain, from energy supply to load scheduling.

[0030] Specifically, the type and intensity of computing tasks, the thermodynamic cycle within the data center, external environmental conditions (such as temperature and humidity), and specific time periods (such as weekdays versus weekends, daytime versus nighttime) all significantly impact energy consumption. Furthermore, data center energy consumption is complexly influenced by factors such as internal hardware configuration, cooling efficiency, and waste heat recovery systems. The interplay between these factors further complicates prediction.

[0031] In the data-driven information age, predicting and managing information load energy consumption faces numerous challenges, particularly the opacity and unpredictable fluctuations of energy consumption. As the core of data processing, data centers consume energy like an iceberg, with enormous hidden costs and environmental burdens, often difficult to accurately control using conventional management methods. This highlights the crucial importance of predicting and managing information load energy consumption.

[0032] Based on the above situation, the present application proposes a method and related device for predicting energy consumption of spatiotemporal scene information load, and the present application is described in detail below with reference to the embodiments and drawings.

[0033] like Figure 1 FIG. 1 is a flow chart of a method for predicting energy consumption of spatiotemporal scene information load in this application, which may include: S101, obtaining computing task feature data, environmental data, historical energy consumption data, spatial data and time stamps in time data to form collected data.

[0034] This step constructs a multimodal dataset that includes computing task characteristics such as CPU utilization and memory usage; environmental parameters such as temperature, humidity, and air quality; historical energy consumption records; spatial layout; and timestamps. The introduction of timestamps enables temporal and spatial alignment of the data, laying the foundation for subsequent dynamic feature extraction. For example, by synchronously recording the server load (85%) at a specific moment in time for instance A and the cooling system power consumption (12kW) in the corresponding area, a correlation map between device-level energy consumption and task intensity can be established.

[0035] S102: performing format unification and encoding processing on the collected data to obtain processed data.

[0036] Because collected data may come from different devices and systems, its data format and units may vary. For example, computing task feature data may be expressed in different units of measurement, and environmental data may contain temperatures in degrees Celsius or Fahrenheit. Format normalization involves converting all data into a unified format and units for subsequent processing and analysis. Non-numeric data, such as device type and computer room location, requires encoding, such as one-hot encoding or label encoding, to convert it into numerical form for computer processing and analysis.

[0037] S103, performing standardization processing on the processed data to obtain standardized data.

[0038] Even after data has been formatted and encoded, its value range and distribution may still vary significantly. The goal of normalization is to transform the data into a standard normal distribution with the same mean and standard deviation, making data with different features comparable. This prevents certain features from dominating the model training due to their large values, thus affecting model accuracy.

[0039] S104: extracting principal components from the standardized data to obtain principal components.

[0040] Normalized data may contain multiple features, some of which may be correlated. Principal component extraction (PCE) is a dimensionality reduction technique that uses linear transformations to convert raw data into a set of independent principal components that maximize the preservation of the original data's information. PCE can reduce data dimensionality and model complexity while removing noise and redundant information, improving model accuracy and generalization.

[0041] S105 , obtaining the current energy consumption based on the principal components and using a multivariate nonlinear regression energy consumption model.

[0042] Based on the extracted principal components, a multivariate nonlinear regression model can be built to calculate current energy consumption. This model considers the combined impact of multiple factors (principal components) on energy consumption and can handle nonlinear relationships between these factors. By training and fitting historical data, the model learns the mapping between different principal components and energy consumption, allowing it to predict current energy consumption based on the current principal component values.

[0043] S106 , obtaining the temperature rise of the data center caused by energy consumption based on the environmental data, computing task characteristic data, and current energy consumption using a thermodynamic calculation model.

[0044] In practical applications, thermodynamic calculation models can be used to calculate the temperature rise in data centers caused by energy consumption based on environmental data, computing task characteristics, and current energy consumption. These models consider factors such as the heat generated by equipment during operation, environmental cooling conditions, and the heat generated by computing tasks. Using principles such as the law of conservation of energy and the heat conduction equation, they calculate the temperature rise in the data center. Temperature rise significantly impacts the performance and lifespan of data center equipment and also feeds back into its energy consumption. Therefore, accurate temperature rise calculation is crucial for energy consumption forecasting and data center management.

[0045] S107: Input the main component, current energy consumption, and temperature rise of the data center caused by energy consumption into an energy consumption prediction model to obtain an energy consumption prediction result.

[0046] In practical applications, energy consumption prediction models can be based on machine learning algorithms, such as neural networks and support vector machines, or time series analysis models, such as ARIMA models. By learning patterns and trends from historical data and combining them with various current feature information, they can predict future energy consumption, providing a basis for decision-making in data center energy management and resource scheduling.

[0047] This application is a strategic method that uses a multivariate nonlinear regression model, combined with spatiotemporal scenario variables and principal component analysis results, to model and predict information load energy consumption. First, principal component analysis is used to extract key feature variables to reduce data dimensions and retain key information. Then, a multivariate nonlinear regression model is used, with principal components and spatiotemporal variables as model inputs, to analyze their impact on energy consumption, and to adjust model parameters based on historical data to improve prediction accuracy. Finally, the relationship between temperature rise and energy consumption of the data center is analyzed in combination with a thermodynamic model, and artificial intelligence technology is used to predict the information load energy consumption values of multiple types of computing tasks, thereby improving the energy efficiency management and operational stability of the data center. It can effectively solve key problems such as inaccurate energy consumption prediction and unreasonable resource allocation faced by current data centers and information technology facilities.

[0048] The following is a more detailed description of the present application through another embodiment of the method for predicting energy consumption of spatiotemporal scene information load. Figure 2 The figure shows the principle diagram of the following embodiment of the spatiotemporal scene information load energy consumption prediction method of the present application.

[0049] S201, data acquisition and processing.

[0050] Data acquisition and processing is responsible for collecting and processing various relevant data to provide a clean, standardized data set for subsequent modeling and prediction. It can be divided into four steps: (1) Data collection.

[0051] Acquire data from multiple sources, including: computing task characteristic data, such as CPU utilization, GPU utilization, memory usage, task execution time, etc.; environmental data, such as the ambient temperature, humidity, server temperature, cooling system parameters of the data center, etc.; historical energy consumption data; spatiotemporal variable data, such as timestamps in temporal data and geographic locations in spatial data.

[0052] (2) Data cleaning.

[0053] Handle missing values, outliers, and inconsistent data to ensure data quality. This may include: Normalize temporal data by converting it into a usable numeric format. Normalize spatial data by converting geographic location data into a unified format, converting latitude and longitude into meters or kilometers for easier calculation. Categorize temporal data and convert categorical variables into numeric variables using one-hot encoding. For example, encode task execution time periods (morning rush hour, noon, evening rush hour, and nighttime) as [1, 0, 0, 0], [0, 1, 0, 0], [0, 0, 1, 0], [0, 0, 0, 1], and so on. Normalize environmental data such as temperature and humidity to ensure consistent data distribution across different features and reduce bias during model training. Perform missing value and 1-only processing on computational task feature data. Specifically, check for missing values in fields such as CPU usage, GPU usage, memory usage, and task execution time. Fill in missing values with low missing values, fill in data with the mean or previous and next values, and mark corresponding type features as unknown. When performing outlier processing, you can identify outliers based on the interquartile range (IQR) or Z-score, correct the outliers to reasonable values, or directly delete the data that is extremely abnormal and cannot be corrected.

[0054] (3) Data standardization.

[0055] The processed data obtained after data cleaning is converted into a standard normal distribution with a mean of 0 and a standard deviation of 1 to eliminate the influence of different feature dimensions. The processed data is standardized so that each variable has a distribution with a mean of 0 and a standard deviation of 1. The standardization formula can be used:

[0056] in, is the processed data, is the mean, is the standard deviation, It is standardized data.

[0057] (4) Principal component extraction.

[0058] Principal Component Analysis (PCA) is used to extract key feature variables, reduce data dimensionality, and retain the most important information. PCA transforms the original data into a new feature space through linear transformation, extracting the features that explain the most data variance. The specific process is as follows:

[0059] in, is the sample size, is the mean vector of the normalized data, is the covariance matrix.

[0060] Pair covariance matrix Perform eigenvalue decomposition to obtain eigenvalues and eigenvectors:

[0061] in, It is a diagonal matrix, and the elements on its diagonal are eigenvalues , is the eigenvector matrix. The eigenvectors corresponding to the largest eigenvalues constitute the principal component matrix By projecting the normalized data onto these eigenvectors, we can obtain the principal components :

[0062] in, yes The matrix represents the front principal components.

[0063] S202, constructing an energy consumption model in the form of multiple nonlinear regression.

[0064] In practical applications, the processed data can be used to construct a multivariate nonlinear regression model to quantify the impact of principal components and spatiotemporal variables on energy consumption. The specific steps are as follows: (1) Construct a multivariate nonlinear regression model.

[0065]

[0066] in, is energy consumption, It is the basic energy consumption level, which is not affected by changes in other variables and reflects the basic energy consumption requirements of the data center. are the coefficients of the principal components, is the principal component variable, are the coefficients of spatiotemporal variables (time, location, temperature), is the spatiotemporal scene variable, is the coefficient of the interaction term, is the error term, which contains random errors and noise and is affected by random fluctuations in the modeling variables and data. is the interaction term between the principal component and the spatiotemporal variables.

[0067] (2) Coefficient estimation and interaction term analysis.

[0068] In this embodiment, the least squares method is used to estimate the coefficients in the regression model. First, the matrix , including spatiotemporal variables and the interaction terms between principal components and spatiotemporal variables .

[0069]

[0070] The two-digit subscript in the above formula is the sample index in the matrix representation, which has the same meaning as the single-letter subscript described in the multivariate nonlinear regression model. The double-letter subscript is the concretization of the matrix. For example, The first spatiotemporal variable (such as temperature) of the sample, The second spatiotemporal variable of the samples (such as humidity).

[0071] Secondly, the coefficient vector is calculated using the least squares formula , which includes 、 and the interaction coefficient :

[0072] The coefficients of the principal components : Each principal component The coefficient of Indicates that the principal component has an impact on energy consumption If If it is positive, it means that the increase of this principal component will lead to an increase in energy consumption; if is negative, indicating that the increase of this principal component will lead to a decrease in energy consumption.

[0073] Coefficients for spatiotemporal variables : Each spatiotemporal variable The coefficient of Indicates that the principal component has an impact on energy consumption If is positive, indicating that the increase of the spatiotemporal variable will lead to an increase in energy consumption; if If it is negative, it means that the increase of this spatiotemporal variable will lead to a decrease in energy consumption. Among them, spatiotemporal variables can include time, geographical location, ambient temperature, etc. These factors directly affect the energy consumption of the data center.

[0074] The coefficients for the interaction terms : Each interaction term The coefficient of Indicates that the principal component has an impact on energy consumption If Is positive, indicating that the principal component and spatiotemporal variables The combined increase of will lead to increased energy consumption; if is negative, indicating that the principal component and spatiotemporal variables The combined increase in energy consumption will result in a reduction in energy consumption.

[0075] In practical applications, the model can be trained using historical energy consumption and principal components as input. The model's performance can then be evaluated using an independent validation dataset to ensure its predictive accuracy and reliability. Feature selection, model regularization, and cross-validation can be used to improve the model's predictive performance and generalization capabilities.

[0076] S203, construct a thermodynamic calculator.

[0077] Thermodynamic calculations, within a spatiotemporal scenario information load energy consumption model based on multivariate nonlinear regression, aim to comprehensively consider the impact of waste heat and ambient temperature on data center energy consumption. Based on the energy consumption of computing tasks, the thermodynamic processes in the data center are analyzed, the temperature rise caused by energy consumption is predicted, and a relationship model between data center temperature rise and energy consumption is established. This can be achieved using the following steps: (1) Data collection.

[0078] The collected data mainly includes three parts: a. Ambient temperature data: Sensors collect environmental parameters such as temperature and humidity inside and outside the data center. b. Equipment operating data: This includes server workload, CPU and GPU temperatures, fan speeds, etc. c. Energy consumption data: The total energy consumption of the entire data center and the energy consumption of each device.

[0079] It should be noted that the equipment working data is the computing task characteristic data collected in step S201. The equipment working data and the ambient temperature can be obtained through step S201, and the energy consumption data can be calculated through step S202.

[0080] (2) Construction of thermodynamic model.

[0081] Calorie calculation: convert energy consumption Converted into heat , assuming that all electrical energy can be converted into heat energy:

[0082] in, The task execution time.

[0083] Heat Balance Equation: Analyze the heat balance of a data center and calculate the net heat change :

[0084] in, It is the heat converted from energy consumption of electricity, wind energy, thermal power plants, etc. is the heat in one part of the cooling system. and All of these can be obtained through the aforementioned heat calculation formula.

[0085] Temperature rise calculation Calculate the temperature rise of the data center based on the heat balance equation and calculate the net heat change :

[0086] in, is the specific heat capacity of air, It's the quality of the air.

[0087] S204: Construct an AI energy consumption predictor as an energy consumption prediction model.

[0088] The AI Energy Consumption Predictor integrates the output of energy consumption models and thermodynamic models with machine learning technology. Its input features include principal components, current energy consumption, and real-time data on the temperature rise of the data center caused by energy consumption. The trained and optimized machine learning model can accurately predict the information load energy consumption of various computing tasks, thereby helping to optimize data center energy efficiency management.

[0089] In practical applications, a random forest regression model can be used as the framework for the AI energy consumption predictor, and the AI energy consumption predictor can be trained in advance. Collect and standardize information load energy consumption data, and train the machine learning model with historical energy consumption data and corresponding feature data. Extract and combine features from historical data and model output to form an input feature vector for training and prediction. Use the random forest regression model algorithm, train the model using historical data, and optimize model parameters through methods such as cross-validation to ensure that the model has good generalization capabilities. Use the trained AI energy consumption predictor to predict energy consumption, using the principal component, current energy consumption, and the temperature rise of the data center caused by energy consumption as input to predict real-time energy consumption. The predicted energy consumption value output by the model is used to monitor and manage the energy efficiency of the data center in real time.

[0090] Large data centers need to accurately predict and manage their energy consumption in order to optimize resource usage and reduce operating costs. This application first cleans and linearly normalizes the data, then standardizes the time and space data, uses principal component analysis to extract characteristic variables, reduces the data dimension, and then constructs a multivariate nonlinear regression model, inputs the principal components and time and space variables, and estimates the coefficients by the least squares method. Energy consumption data is analyzed using thermodynamic calculations, and a relationship model between temperature rise and energy consumption is established. Finally, a random forest regression model is used to train the data, and the model parameters are optimized through cross-validation. The principal components, current energy consumption, and temperature rise caused by energy consumption in the data center are input to predict real-time energy consumption. The output results are used to monitor and manage the energy efficiency of the data center, optimize resource usage, and reduce operating costs.

[0091] like Figure 3 FIG. 1 is a schematic diagram of a spatiotemporal scene information load energy consumption prediction system of the present application, which may include: The data acquisition module is used to obtain computing task feature data, environmental data, historical energy consumption data, spatial data, and time stamps in time data to form collected data; A data processing module, configured to perform format unification and encoding processing on the collected data to obtain processed data; A standardization module, configured to perform standardization on the processed data to obtain standardized data; A principal component extraction module is used to extract principal components from the standardized data to obtain principal components; An energy consumption calculation module, configured to obtain current energy consumption based on the principal components using an energy consumption model in the form of multiple nonlinear regression; The temperature rise calculation module is used to calculate the temperature rise caused by energy consumption in the data center based on environmental data, computing task characteristic data, and current energy consumption using a thermodynamic calculation model. The prediction module is used to input the main components, current energy consumption, and temperature rise of the data center caused by energy consumption into the energy consumption prediction model to obtain energy consumption prediction results.

[0092] It should be noted that in the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the system embodiments described above are merely schematic. For example, the division of each module is only a logical function division. There may be other division methods in actual implementation. For example, multiple modules can be combined or integrated into another device, or some features can be ignored or not executed. The modules described as separate components may or may not be physically separated. The components displayed as modules may be one physical unit or multiple physical units, that is, they may be located in one place, or they may be distributed in multiple different places. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0093] In addition, the modules in the various embodiments of the present invention may be integrated into a single processing unit, each module may exist physically separately, or two or more modules may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0094] An embodiment of the present application also provides an electronic device, which may include one or more processors, memories, and communication interfaces.

[0095] The memory, the communication interface, and the processor are coupled together. For example, the memory, the communication interface, and the processor may be coupled together via a bus.

[0096] The communication interface is used to transmit data with other devices. The memory stores computer program code. The computer program code includes computer instructions that, when executed by the processor, cause the electronic device to perform the steps of the above-mentioned spatiotemporal scenario information load energy consumption prediction method.

[0097] Among them, the processor can be a processor or a controller, for example, a central processing unit (CPU), a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It can implement or execute the various exemplary logic blocks, modules and circuits described in conjunction with the contents of this disclosure. The processor can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, and the like. The processor can be used to support electronic devices in executing the method steps provided in the above embodiments.

[0098] The bus may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The above buses may be divided into an address bus, a data bus, a control bus, etc.

[0099] An embodiment of the present application provides a computer-readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned spatiotemporal scene information load energy consumption prediction method are implemented.

[0100] The computer-readable storage medium involved in this application includes random access memory (RAM), internal memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disks, removable disks, CD ROMs, or any other form of storage medium known in the technical field.

[0101] The above are merely preferred embodiments of the present application and are not intended to limit the present application. Those skilled in the art will readily appreciate that various modifications and variations are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present application shall be included within the scope of protection of the present application.

Claims

1. A method for predicting energy consumption of spatiotemporal scene information load, characterized in that: include: Obtain computing task feature data, environmental data, historical energy consumption data, spatial data, and time stamps from time data to form collected data; Performing format unification and coding processing on the collected data to obtain processed data; performing standardization on the processed data to obtain standardized data; Performing principal component extraction on the standardized data to obtain principal components; According to the principal components, the current energy consumption is obtained by using an energy consumption model in the form of multiple nonlinear regression; Based on environmental data, computing task characteristic data, and current energy consumption, the temperature rise of the data center caused by energy consumption is obtained using a thermodynamic calculation model. The main components, current energy consumption, and temperature rise of the data center caused by energy consumption are input into the energy consumption prediction model to obtain the energy consumption prediction results.

2. The method for predicting energy consumption of spatiotemporal scene information load according to claim 1, characterized in that: The method for formatting and encoding the collected data includes: Classify the timestamps in time data and convert them into numerical format; Carry out coding and standardization processing on spatial data; Perform data distribution consistency processing on environmental data; Process missing values and outliers in computing task feature data.

3. The method for predicting energy consumption of spatiotemporal scene information load according to claim 1, characterized in that: The method for standardizing the processed data comprises: The processed data were transformed into a standard normal distribution.

4. The method for predicting energy consumption of spatiotemporal scene information load according to claim 1, characterized in that: The method for extracting principal components from the standardized data comprises: The covariance matrix is obtained by the following formula : in, is the sample size, For the Standardized data, is the mean vector of the standardized data; The covariance matrix is expressed as follows Perform eigenvalue decomposition: in, is the eigenvector matrix, is a diagonal matrix; The elements on the diagonal of the diagonal matrix are used as eigenvalues, and the eigenvectors corresponding to the eigenvalues with the largest preset number are selected to form the principal component matrix ; Project the normalized data onto the principal component matrix The principal components are obtained from the eigenvectors of .

5. The method for predicting energy consumption of spatiotemporal scene information load according to claim 1, characterized in that: The expression of the energy consumption model in the form of multiple nonlinear regression includes: in, is energy consumption, As the basic energy consumption level, are the coefficients of the principal components, is the principal component variable, is the coefficient of the spatiotemporal variable, is the spatiotemporal scene variable, is the coefficient of the interaction term, is the error term.

6. The method for predicting energy consumption of spatiotemporal scene information load according to claim 1, characterized in that: The expression of the thermodynamic calculation model includes: in, The temperature rise caused by energy consumption in the data center, is the net heat change, is the specific heat capacity of air, For the quality of air.

7. The method for predicting energy consumption of spatiotemporal scene information load according to claim 1, characterized in that: The energy consumption prediction model adopts a random forest regression model.

8. A spatiotemporal scene information load energy consumption prediction system, characterized in that: include: The data acquisition module is used to obtain computing task feature data, environmental data, historical energy consumption data, spatial data, and time stamps in time data to form collected data; A data processing module, configured to perform format unification and encoding processing on the collected data to obtain processed data; A standardization module, configured to perform standardization on the processed data to obtain standardized data; A principal component extraction module is used to extract principal components from the standardized data to obtain principal components; an energy consumption calculation module, configured to obtain current energy consumption based on the principal components using an energy consumption model in the form of multiple nonlinear regression; The temperature rise calculation module is used to calculate the temperature rise caused by energy consumption in the data center based on environmental data, computing task characteristic data, and current energy consumption using a thermodynamic calculation model. The prediction module is used to input the main components, current energy consumption, and temperature rise of the data center caused by energy consumption into the energy consumption prediction model to obtain energy consumption prediction results.

9. An electronic device, characterized in that: include: A memory, one or more processors; the memory is coupled to the processor; wherein computer program code is stored in the memory, and the computer program code includes computer instructions, and when the computer instructions are executed by the processor, the electronic device executes the steps of the spatiotemporal scene information load energy consumption prediction method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the spatiotemporal scene information load energy consumption prediction method according to any one of claims 1 to 7 are implemented.

Images