Energy consumption management method and device and nonvolatile storage medium

Abstract

The invention discloses an energy consumption management method and device and a nonvolatile storage medium. The method comprises the following steps: acquiring related data of candidate equipment in a target scene; feature extraction is carried out on the related data to obtain a multi-dimensional feature vector, and the multi-dimensional feature vector is used for describing multi-dimensional features related to the energy consumption of the candidate equipment; an energy consumption prediction model is adopted to process and analyze the multi-dimensional feature vector, an energy consumption prediction value output by the energy consumption prediction model is obtained, and the energy consumption prediction model comprises a time sequence prediction model and a deep learning model with output of the time sequence prediction model as input; and determining a comprehensive score of each candidate device according to the energy consumption predicted value and the historical energy consumption value pair, and managing the energy consumption of the target scene according to the comprehensive score and the indication information. The technical problems of long prediction time and low prediction efficiency when a deep neural network model is independently adopted for energy prediction are solved.

Description

Technical Field

[0001] This application relates to the field of energy management technology, and more specifically, to an energy consumption management method and apparatus, and a non-volatile storage medium. Background Technology

[0002] In related technologies, energy prediction directly based on deep neural networks faces challenges. Training deep neural networks typically requires significant computing resources (GPUs) and considerable time, especially when dealing with datasets containing multiple features and complex relationships. Furthermore, the selection of model parameters (such as network architecture, learning rate, and batch size) may necessitate extensive experimentation and tuning, increasing the cost and complexity of model development. Similarly, using trained deep neural networks for energy consumption prediction also requires processing datasets with multiple features and complex relationships, consuming substantial computing resources and resulting in long prediction times and low prediction efficiency.

[0003] There is currently no effective solution to the above problems. Summary of the Invention

[0004] This application provides an energy management method and apparatus, and a non-volatile storage medium, to at least solve the technical problems of long prediction time and low prediction efficiency when using a deep neural network model alone for energy prediction.

[0005] According to one aspect of the embodiments of this application, an energy consumption management method is provided, comprising: acquiring relevant data of candidate devices in a target scenario, wherein the relevant data includes: energy consumption data generated by the candidate devices during the period to be tested, the target scenario includes: a computer room, a data center, and the candidate devices include: communication devices, and basic equipment related to the operating environment of the communication devices; extracting features from the relevant data to obtain multi-dimensional feature vectors, wherein the multi-dimensional feature vectors are used to describe multi-dimensional features related to the energy consumption of the candidate devices; processing and analyzing the multi-dimensional feature vectors using an energy consumption prediction model to obtain an energy consumption prediction value output by the energy consumption prediction model, wherein the energy consumption prediction value is used to indicate the energy consumption generated by the candidate devices during future periods, and the energy consumption prediction model includes: a time series prediction model and a deep learning model with the output of the time series prediction model as input; determining a comprehensive score for each candidate device based on the energy consumption prediction value and historical energy consumption value, and managing the energy consumption of the target scenario based on the comprehensive score and indication information.

[0006] Optionally, the relevant data also includes: basic equipment data for describing the rated information of the equipment, and environmental data for describing the operating environment of the candidate equipment; feature extraction is performed on the relevant data to obtain multi-dimensional feature vectors, including: feature extraction is performed on different types of relevant data from different feature extraction dimensions to obtain multiple independent feature vectors, wherein the independent feature vectors include: equipment feature vectors obtained by feature extraction of basic equipment data, environmental feature vectors obtained by feature extraction of environmental data, and energy consumption feature vectors obtained by feature extraction of energy consumption data; multiple independent feature vectors are concatenated into a multi-dimensional feature vector, wherein the number of columns of the multi-dimensional feature vector is greater than the number of columns of any single independent feature vector.

[0007] Optionally, before processing and analyzing the multi-dimensional feature vector using the energy consumption prediction model, the process includes: determining the target elements related to energy consumption in the multi-dimensional feature vector, and combining multiple target columns to which multiple target elements belong into a target feature vector; encoding the non-numerical elements in the target feature vector into numerical form, and normalizing the numerical elements in the target feature vector.

[0008] Optionally, the energy consumption prediction model is used to process and analyze the multi-dimensional feature vectors to obtain the energy consumption prediction value output by the energy consumption prediction model, including: using the time series prediction model in the energy consumption prediction model to process and analyze the multi-dimensional feature vectors and output the energy consumption change characteristics, wherein the energy consumption change characteristics include: the period of energy consumption change and the energy consumption change trend in each period; using the deep learning model in the energy consumption prediction model to process the energy consumption change characteristics and output the energy consumption prediction value.

[0009] Optionally, the energy consumption prediction model is trained using the following method: The historical time series stored in the distributed storage system and the corresponding historical energy consumption series are determined as the first type of training data, and the time series prediction model is trained using this first type of training data. The time series records the historical operating times of candidate devices, and the historical energy consumption series records the historical energy consumption generated by the candidate devices during those historical operating times. Once the time series prediction model training is complete, the first type of training data and the output of the time series prediction model for the first type of training data are determined as the second type of training data. The deep learning model is iteratively trained using the second type of training data. During iterative training, the training progress is determined based on the change in the loss function and the preset number of iterations. The training progress includes: continuing training and stopping training.

[0010] Optionally, a comprehensive score for each candidate device is determined based on the predicted energy consumption and historical energy consumption, including: for each candidate device, determining an energy efficiency score based on the rated energy consumption and predicted energy consumption; determining a reliability score based on the predicted energy consumption and historical energy consumption; determining a cost-effectiveness score based on the predicted energy consumption, unit energy cost, and other preset costs; and determining a comprehensive score based on the energy efficiency score, a first weight of the energy efficiency score, a reliability score, a second weight of the reliability score, a cost-effectiveness score, and a third weight of the cost-effectiveness score.

[0011] Optionally, managing the energy consumption of the target scenario based on the comprehensive score and instruction information includes: receiving instruction information, wherein the instruction information is used to instruct the reconfiguration of candidate devices in the target scenario; generating an optimal configuration scheme based on the energy consumption constraints, multiple candidate devices, and multiple comprehensive scores recorded in the instruction information, and implementing the optimal configuration scheme on the target scenario, wherein, when the optimal configuration scheme is implemented on the target scenario, the energy consumption generated by the target scenario satisfies the energy consumption constraints, and the optimal configuration scheme includes at least one candidate device.

[0012] According to another aspect of the embodiments of this application, an energy consumption management device is also provided, comprising: an acquisition module, configured to acquire relevant data of candidate devices in a target scenario, wherein the relevant data includes: energy consumption data generated by the candidate devices during the period to be detected, the target scenario includes: a computer room, a data center, and the candidate devices include: communication equipment, and basic equipment related to the operating environment of the communication equipment; a processing module, configured to extract features from the relevant data to obtain a multi-dimensional feature vector, wherein the multi-dimensional feature vector is used to describe multi-dimensional features related to the energy consumption of the candidate devices; a prediction module, configured to process and analyze the multi-dimensional feature vector using an energy consumption prediction model to obtain an energy consumption prediction value output by the energy consumption prediction model, wherein the energy consumption prediction value is used to indicate the energy consumption generated by the candidate devices during future periods, and the energy consumption prediction model includes: a time series prediction model and a deep learning model with the output of the time series prediction model as input; and a device management module, configured to determine a comprehensive score for each candidate device based on the energy consumption prediction value and historical energy consumption value, and manage the energy consumption of the target scenario based on the comprehensive score and indication information.

[0013] According to another aspect of the embodiments of this application, a non-volatile storage medium is also provided, in which a computer program is stored, wherein the above-described energy management method is executed by running the computer program in the device where the non-volatile storage medium is located.

[0014] According to another aspect of the embodiments of this application, an electronic device is also provided, including a memory and a processor, wherein the memory stores a computer program and the processor is configured to execute the above-described energy management method through the computer program.

[0015] According to another aspect of the embodiments of this application, a computer program product is also provided, including computer instructions, which, when executed by a processor, implement the steps of the above-described energy management method.

[0016] In this embodiment, relevant data of candidate devices in a target scenario is acquired. This data includes energy consumption data generated by the candidate devices during the detection period. The target scenario includes a computer room and a data center. The candidate devices include communication equipment and basic equipment related to the operating environment of the communication equipment. Feature extraction is performed on the relevant data to obtain multi-dimensional feature vectors, which describe multi-dimensional features related to the energy consumption of the candidate devices. An energy consumption prediction model is used to process and analyze the multi-dimensional feature vectors to obtain the energy consumption prediction value output by the model. This predicted value indicates the energy consumption generated by the candidate devices during future periods. The energy consumption prediction model includes time series prediction. The model combines a time-series prediction model with a deep learning model as input. It determines a comprehensive score for each candidate device based on predicted and historical energy consumption values, and manages energy consumption in target scenarios based on this comprehensive score and indicator information. By employing a prediction model jointly trained by the time-series prediction model and the deep learning model, energy consumption prediction is achieved. The time-series prediction model captures the periodicity and trends of energy consumption, reducing the steps required for the deep learning model to extract time-series features during model application, thus shortening the time required for energy consumption prediction and improving its efficiency. This solves the technical problems of long prediction time and low prediction efficiency inherent in using deep neural network models alone for energy prediction. Furthermore, the deep learning model's in-depth analysis of complex data relationships significantly improves prediction accuracy. By comprehensively evaluating the energy efficiency, reliability, and cost-effectiveness of equipment, it provides a scientific basis for equipment selection in computer rooms and data centers, effectively reducing energy costs and improving resource utilization. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0018] Figure 1 This is a hardware structure block diagram of a computer terminal for implementing an energy management method according to an embodiment of this application;

[0019] Figure 2 This is a flowchart of the steps of an energy management method according to an embodiment of this application;

[0020] Figure 3 This is a structural diagram of an energy management device according to an embodiment of this application. Detailed Implementation

[0021] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0022] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0023] To better understand the embodiments of this application, the technical terms involved in the embodiments of this application are explained below:

[0024] One-hot encoding: a technique that converts categorical features into numerical features, which are then converted into binary vectors.

[0025] In related technologies, the aim is to predict energy consumption through machine learning. The main technical details are as follows: Data is collected from multiple sources, including information systems, IoT network systems, and environmental data such as temperature and humidity. During data preprocessing, methods such as outlier elimination, missing value replacement, and variable reduction are employed. Feature engineering is performed on the preprocessed data, considering both temporal features (e.g., month, season, weekday / weekend) and environmental features (e.g., temperature and humidity). A deep neural network is trained based on both temporal and environmental features; that is, the deep neural network must simultaneously learn the temporal and environmental characteristics of energy consumption. Therefore, there are problems such as long training time, high training cost (computational resource cost), and long time required to use the trained deep neural network for energy consumption prediction. To address this problem, this application provides a related solution, which is detailed below.

[0026] According to an embodiment of this application, an embodiment of an energy management method is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0027] The methods and embodiments provided in this application can be executed on mobile terminals, computer terminals, or similar computing devices. Figure 1 A hardware block diagram of a computer terminal for implementing an energy management method is shown. Figure 1 As shown, the computer terminal 10 may include one or more processors 102 (shown as 102a, 102b, ..., 102n in the figure) 102 (processor 102 may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.), a memory 104 for storing data, and a transmission device 106 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of a BUS bus), a network interface, a power supply, and / or a camera. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, computer terminal 10 may also include... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0028] It should be noted that the aforementioned one or more processors 102 and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be embodied, in whole or in part, in software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuits may be a single, independent processing module, or may be integrated, in whole or in part, into any other element within the computer terminal 10. As involved in the embodiments of this application, the data processing circuits serve as a form of processor control (e.g., selection of a variable resistor termination path connected to an interface).

[0029] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the energy management method in this embodiment. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, thereby realizing the aforementioned energy management method. The memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the computer terminal 10 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0030] The transmission device 106 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of the computer terminal 10. In one example, the transmission device 106 includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 106 may be a Radio Frequency (RF) module, used for wireless communication with the Internet.

[0031] The display may be, for example, a touchscreen liquid crystal display (LCD) that allows the user to interact with the user interface of the computer terminal 10.

[0032] This application provides an energy management method that can operate under the above-described operating environment. Figure 2 This is a flowchart of the steps of the energy management method provided in the embodiments of this application, such as... Figure 2 As shown, the method includes the following steps:

[0033] Step S202: Obtain relevant data of candidate devices in the target scenario. The relevant data includes: energy consumption data generated by the candidate devices during the testing period. The target scenario includes: computer room and data center. The candidate devices include: communication equipment and basic equipment related to the operating environment of the communication equipment.

[0034] The method provided in this application embodiment can be applied to high-device-density scenarios (i.e., target scenarios) such as data centers and large computer rooms. By collecting energy consumption data of devices (i.e., candidate devices) within the target scenario and combining it with a predictive model, it provides decision support for the selection of new equipment and the optimization of existing equipment, achieving refined energy consumption management, improving energy utilization efficiency, and reducing operating costs. In step S202, relevant data of all candidate devices within the target scenario (such as computer rooms and data centers) are collected. For example, energy consumption data generated by all candidate devices during the testing period is collected. The aforementioned candidate devices include not only communication equipment (such as servers, routers, switches, etc.) but also basic equipment closely related to the operating environment of communication equipment (such as temperature control equipment, lighting equipment, etc.). The data acquired in step S202 can be stored in a distributed storage system, such as a distributed big data storage platform built on a framework based on large-scale datasets (Hadoop). This distributed storage platform / system utilizes the Hadoop Distributed File System (HDFS) to achieve highly reliable and high-throughput data storage, provides data warehouse functionality through data warehouse tools (Hive) to support data analysis, and enables real-time random data read / write through a database (HBase). The database defines fields such as DeviceData, DeviceType, Model, Configuration, RatedPower, Energy Efficiency Ratio (PUE), Power Curve, and Environment Data to store the relevant data acquired above. For data that cannot be directly obtained, manual entry can be used to ensure data integrity.

[0035] Step S204: Extract features from the relevant data to obtain a multi-dimensional feature vector, wherein the multi-dimensional feature vector is used to describe the multi-dimensional features related to the energy consumption of the candidate device.

[0036] After collecting relevant data from candidate devices, in step S204, feature extraction is performed on this data to form a multi-dimensional feature vector that can describe the energy consumption behavior of the candidate devices. This process aims to convert the raw data into a format that the model can understand for subsequent energy consumption prediction.

[0037] Optionally, the relevant data also includes: basic equipment data for describing the rated information of the equipment, and environmental data for describing the operating environment of the candidate equipment; feature extraction is performed on the relevant data to obtain multi-dimensional feature vectors, including: feature extraction is performed on different types of relevant data from different feature extraction dimensions to obtain multiple independent feature vectors, wherein the independent feature vectors include: equipment feature vectors obtained by feature extraction of basic equipment data, environmental feature vectors obtained by feature extraction of environmental data, and energy consumption feature vectors obtained by feature extraction of energy consumption data; multiple independent feature vectors are concatenated into a multi-dimensional feature vector, wherein the number of columns of the multi-dimensional feature vector is greater than the number of columns of any single independent feature vector.

[0038] The relevant data for candidate devices can be provided by the candidate device manufacturers. In addition to providing energy consumption data generated during the operation of the candidate devices, these manufacturers also provide key information such as device type, model, configuration, rated power, and power usage efficiency (PUE). Among these, device type, model, configuration, rated power, and PUE are basic device data used to describe the fundamental attributes of the candidate devices. Basic equipment in the target scenario includes temperature control devices (such as air conditioners). These devices regulate the temperature of the operating environment of the candidate devices, and the temperature of the operating environment affects the energy consumption of the communication devices. Therefore, the relevant data obtained in step S202 also includes environmental data. Based on the above, the relevant data for candidate devices contains various types of data. Therefore, in this embodiment, when performing data feature extraction, features are extracted from different feature dimensions for each type of data to obtain a vector (i.e., an independent feature vector) describing one dimension of the candidate device's features. The process of extracting independent feature vectors is as follows: Based on the needs of the time series prediction model, key features are extracted from relevant data, including: extracting time features from energy consumption data, such as hour, date, month, periodicity, and holidays, to generate time feature vectors; these time feature vectors are used to analyze the relationship between energy consumption and time. Environmental features are extracted from environmental data, such as temperature and humidity, to generate environmental feature vectors. Equipment features are extracted from basic equipment data, such as equipment model, configuration parameters, rated power, and energy efficiency ratio, to generate equipment feature vectors. Historical energy consumption features are extracted from historical energy consumption data, such as historical energy consumption values ​​and energy consumption trends. By calculating the standard deviation of historical energy consumption, the time series prediction model can learn the patterns and trends of energy consumption changes. After feature extraction, the above independent feature vectors (equipment feature vector, environmental feature vector, and energy consumption feature vector) are horizontally concatenated to form a multi-dimensional feature vector; this multi-dimensional feature vector contains all relevant equipment information, providing rich input for model training. Horizontal concatenation refers to concatenating feature vectors along the row direction. Therefore, the number of columns in the multi-dimensional feature vector generated by concatenation will be greater than the number of columns in each independent feature vector that generated the multi-dimensional feature vector.

[0039] According to some optional embodiments of this application, before processing and analyzing the multi-dimensional feature vector using the energy consumption prediction model, the method includes: determining the target elements related to energy consumption in the multi-dimensional feature vector, and combining the multiple target columns to which the multiple target elements belong into a target feature vector; encoding the non-numerical elements in the target feature vector into numerical form, and normalizing the numerical elements in the target feature vector.

[0040] In this embodiment, to reduce the complexity of data preprocessing, the data preprocessing process before training the energy consumption prediction model is as follows: First, for the fused feature vector generated after feature extraction and concatenation, a filtering method is used to reduce the dimensionality of the multi-dimensional feature vector to obtain the dimensionality reduction result (i.e., the target feature vector), thereby reducing the amount of data processing during energy consumption prediction and avoiding invalid calculations. Specifically, columns useful for energy consumption prediction (i.e., target columns) are identified in the multi-dimensional feature vector. For example, elements representing energy consumption-related features and elements representing time features are identified as energy consumption-related elements (i.e., target elements). Columns containing these target elements in the multi-dimensional feature vector will be identified as target columns. These target columns are extracted to form the target feature vector. The target feature vector may contain non-numerical elements, such as equipment model, weekday / weekend categories, etc. To input these features into the energy consumption prediction model based on numerical computation, the non-numerical elements need to be encoded into numerical form. For example, one-hot encoding is used to encode non-numerical elements into numerical form, converting each non-numerical element into a binary vector. For numerical elements, such as categorical features like temperature, humidity, and rated power, normalization is needed to eliminate the influence of different dimensions and improve prediction accuracy. Normalization aims to scale feature values ​​to a fixed range, for example, scaling them to the [0,1] interval. For instance, if the energy consumption range of a device is well-defined (e.g., rated power ≤ 1000 watts), normalization can normalize it to the numerical range [0,1]. Normalization can be implemented in various ways, such as min-max scaling (MinMaxScaler) or standard deviation scaling (StandardScaler). Finally, the standardized feature vectors are combined with the target variable (e.g., actual energy consumption) from the original data to obtain the preprocessed dataset.

[0041] Step S206: The multi-dimensional feature vector is processed and analyzed using the energy consumption prediction model to obtain the energy consumption prediction value output by the energy consumption prediction model. The energy consumption prediction value is used to indicate the energy consumption generated by the candidate device during operation in the future. The energy consumption prediction model includes a time series prediction model and a deep learning model that takes the output of the time series prediction model as input.

[0042] In step S206, the energy consumption prediction model is used to process and analyze the multi-dimensional feature vector obtained in step S204 to predict the energy consumption generated by the candidate device during future periods. The future periods can be periods after the current time, such as one day after the current time. The energy consumption prediction model is a hybrid model that combines a time series prediction model (such as Prophet) and a deep learning model (such as TensorFlow). When processing and analyzing the multi-dimensional feature vector, the output of the time series prediction model will be used as one of the input data of the deep learning model.

[0043] In step S206, the energy consumption prediction model can be loaded into memory. For example, the raw data of the energy consumption prediction model can be loaded from non-volatile memory into volatile memory so that the processor can run the energy consumption prediction model. The raw data of the energy consumption prediction model refers to unprocessed data, which typically includes the parameters and structural data of the energy consumption prediction model. The structural data can be the calculation relationships based on the parameters, such as the forward propagation calculation relationships between intermediate layers and between neurons. Specifically, the structural data can include the structure-related code of the energy consumption prediction model, such as code used to perform related calculations between intermediate layers and between neurons.

[0044] In one implementation, a region can be partitioned in memory for loading the energy consumption prediction model, which may include a structure data storage region and a parameter storage region. The structure data storage region stores structure-related code, and the parameters referenced by it can be accessed via pointers pointing to the addresses of specific parameters in the parameter storage region. During the training of the energy consumption prediction model, frequent parameter updates may be required; in this case, updating the parameter values ​​in the parameter storage region is sufficient.

[0045] Optionally, the energy consumption prediction model is used to process and analyze the multi-dimensional feature vectors to obtain the energy consumption prediction value output by the energy consumption prediction model, including: using the time series prediction model in the energy consumption prediction model to process and analyze the multi-dimensional feature vectors and output the energy consumption change characteristics, wherein the energy consumption change characteristics include: the period of energy consumption change and the energy consumption change trend in each period; using the deep learning model in the energy consumption prediction model to process the energy consumption change characteristics and output the energy consumption prediction value.

[0046] As mentioned in the above embodiments, the energy consumption prediction model is a hybrid model constructed from a time series prediction model and a deep learning model. The time series prediction model serves as the front end of the energy consumption prediction model, responsible for processing and extracting time series features. Specifically, the time series prediction model performs time series analysis on multi-dimensional feature vectors to capture energy consumption change trends, seasonality, and periodicity of energy consumption changes in candidate devices. The deep learning model is used to further explore the complex relationships between device energy consumption data and other data. When initializing the time series prediction model, the prior scale of the trend change point is set to 0.1, and the prediction interval width (interval_width) is set to 0.95 to ensure that the time series prediction model captures a wider range of uncertainties. Simultaneously, multiple seasonal configurations (annual / weekly / daily cycles) are enabled to better match the actual changing patterns of energy consumption data. Specifically, when analyzing annual fluctuation details, annual cycle analysis is performed, with the corresponding Fourier series (fourier_order) set to 10; when differentiating energy consumption between weekdays and weekends, weekly cycle analysis is performed, with the corresponding Fourier series (fourier_order) set to 5; and when capturing daily energy consumption peaks, daily cycle analysis is performed, with the corresponding Fourier series (fourier_order) set to 3. When applying the predictive model for energy consumption forecasting, multi-dimensional feature vectors, including timestamps and energy consumption values, are input into the time series prediction model. The time series prediction model automatically analyzes trend, seasonality, and cyclical characteristics. The output of the time series prediction model serves as one of the input features for the deep learning model. When initializing a deep learning model, a machine learning framework (TensorFlow) is used to create a deep learning model containing neural network layers (LSTM layers) and fully connected layers (Dense layers). For example, to create a deep learning model with two neural network layers, the creation process is as follows: First, add a first LSTM layer containing 50 neurons. The input shape is (number of features, 1), where the number of features is the total number of fused feature vectors, and 1 represents the data dimension of a single time step. Add a dropout layer with a dropout rate of 0.2 to prevent overfitting. Next, add a second LSTM layer, also containing 50 neurons, and add a dropout layer with a dropout rate of 0.2. Finally, add a fully connected layer to output the final energy consumption prediction. The fully connected layer performs a linear transformation, which can be expressed as Y = XW + B, where X is the input of the deep learning model, W is the weight matrix of the deep learning model, B is the bias vector, and Y is the output energy consumption prediction.

[0047] When creating a deep learning model, training parameters are also configured for it: an optimizer is defined, the learning rate is set to 0.001, the weight decay is (5e-4), and the mean squared error (MSE) is used as the loss function.

[0048] According to some optional embodiments of this application, the energy consumption prediction model is trained by the following method: The historical time series stored in the distributed storage system and the corresponding historical energy consumption sequence are determined as the first type of training data, and the time series prediction model is trained using the first type of training data. The time series records the historical operating times of candidate devices, and the historical energy consumption sequence records the historical energy consumption generated by the candidate devices during those historical operating times. When the time series prediction model training is determined to be complete, the first type of training data and the output of the time series prediction model for the first type of training data are determined as the second type of training data. The deep learning model is iteratively trained using the second type of training data. During the iterative training process, the training progress is determined based on the change in the loss function and the preset number of iterations. The training progress includes: continuing training and stopping training.

[0049] The training process of the energy consumption prediction model used in this embodiment is as follows: Historical operating times (time series) and corresponding energy consumption values ​​(historical energy consumption series) of devices are extracted from the distributed storage system. The time series should include precise dates and times, while the historical energy consumption series represents the actual energy consumption values ​​of the devices at these time points. The extracted data will be used as training data for the time series model (i.e., the first type of training data). During the training of the energy consumption prediction model, the time series prediction model is trained separately first. The first type of training data is input into the time series prediction model to learn the characteristics and patterns of the time series. Parameter adjustments, such as seasonality and trends, are made to adapt to the data characteristics. After the time series prediction model is trained, the output of the time series prediction model and other auxiliary features (such as time features, environmental features, etc., data included in the first type of training data) are combined as training data for the deep learning model (i.e., the second type of training data). The deep learning model is then trained using this training data. In this embodiment, an early stopping function is added during the training of the deep learning model. This function is implemented using both the loss function and a preset number of training epochs. Specifically, during training, the model's loss function value is monitored. If the change in the loss function does not improve over several consecutive training epochs (e.g., 5), training is stopped early, even before the preset number of training epochs is reached, to prevent overfitting. The change in the loss function is the difference between the two loss functions corresponding to two adjacent iterations. For example, if the number of training epochs is set to 100, allowing the model to perform 100 complete forward and backward propagations on the entire training dataset, but after the 80th iteration, the change in the loss function does not improve over 5 consecutive training epochs, training is stopped early. Otherwise, if the change in the loss function continues to improve and the preset number of iterations is not reached, the deep learning model continues to be trained iteratively. When training a deep learning model, a batch size is typically set, dividing all the training data of the deep learning model into multiple batches for processing, thereby improving the training efficiency of the model.

[0050] During model training, the following methods can be used to evaluate and optimize the energy consumption prediction model. First, after preprocessing and feature extraction of the data in the distributed storage system, multiple fused feature vectors are generated. These fused feature vectors are then divided into three parts according to a preset ratio: training set, validation set, and test set. For example, 70% is the training set, 15% is the validation set, and 15% is the test set. During model evaluation, the test set data is used. The prediction accuracy of the energy consumption prediction model is evaluated by calculating the mean squared error (MSE) between the predicted and actual values. Using the test set data, which is data not used during model training, can more objectively reflect the model's generalization ability. For example, if the target variable (energy consumption) is approximately 100-1000 watts, after normalization and scaling to [0,1], a mean squared error (MSE) < 0.01 indicates good model performance, reaching a practical accuracy level. The formula for calculating the mean squared error (MSE) is: Where n is the total number of samples in the dataset (e.g., n = 100 if there are 100 data points), y1 represents the true value, i.e., the actual energy consumption at a certain historical moment, and y2 is the predicted value, i.e., the predicted energy consumption of the model for a certain historical moment during training. Based on the model evaluation results, the model is optimized to improve its predictive performance. The optimization trigger condition is when the model evaluation metric does not meet the expected target, such as an excessively large MSE (e.g., assuming a good MSE threshold of 100 kilowatts, an MSE value greater than 100 kilowatts is too large), requiring model optimization. In this case, hyperparameter tuning is performed using Bayesian optimization. By setting the number of iterations to 50, using the MSE on the validation set as the objective function, parameters such as the learning rate, batch size, dropout rate, and network structure parameters (number of units in LSTM layers, number of units in Dense layers) are adjusted. The model is retrained using the adjusted parameters, and then evaluated using test set data to compare the evaluation metrics before and after optimization. Finally, based on the results of the re-evaluation, repeat the steps of parameter adjustment, retraining, and re-evaluation until the MSE range is <0.01 and the performance of the energy consumption prediction model reaches a satisfactory level.

[0051] Step S208: Determine the comprehensive score of each candidate device based on the predicted energy consumption value and the historical energy consumption value, and manage the energy consumption of the target scenario based on the comprehensive score and indication information.

[0052] After obtaining the predicted energy consumption value for each candidate device in step S206, in step S208, a comprehensive evaluation of each candidate device is performed by combining its historical energy consumption data (energy consumption generated by the candidate device before the current moment) to determine its overall score. The overall score of the candidate device is obtained by comprehensively evaluating the candidate device based on energy consumption and other energy-related dimensions (such as operating status and operating cost). After determining the overall score of each candidate device, when the user instructs energy consumption management for the target scenario and provides specific energy consumption management requirements (both the instruction and the specific requirements are recorded in the instruction information), the overall score of the candidate device and the instruction information are combined to achieve energy consumption management for the target scenario.

[0053] According to some optional embodiments of this application, a comprehensive score for each candidate device is determined based on predicted energy consumption and historical energy consumption, including: for each candidate device, determining an energy efficiency score based on the rated energy consumption and predicted energy consumption; determining a reliability score based on predicted energy consumption and historical energy consumption; determining a cost-effectiveness score based on predicted energy consumption, unit energy consumption cost, and other preset costs; and determining a comprehensive score based on the energy efficiency score, a first weight of the energy efficiency score, a reliability score, a second weight of the reliability score, a cost-effectiveness score, and a third weight of the cost-effectiveness score.

[0054] After obtaining the predicted energy consumption values ​​of candidate equipment, the predicted energy consumption values ​​output by the energy consumption prediction model and historical energy consumption data (i.e., energy consumption data generated by the candidate equipment during operation before the current moment) are combined to comprehensively evaluate all candidate equipment. Then, the best equipment solution is selected based on the score, providing decision support for the equipment construction project phase. The comprehensive score is determined jointly based on the energy efficiency score, reliability score, and cost-effectiveness score. Among them, the energy efficiency score = energy efficiency ratio = (rated power / predicted average energy consumption) * 100%. The higher the value, the higher the energy efficiency of the equipment. The rated power (i.e., rated energy consumption) is recorded in the relevant data of the candidate equipment. The predicted average energy consumption is the ratio of the predicted energy consumption value of the candidate equipment in the future period to the energy consumption of the future period (such as hourly, daily, or monthly). Reliability score = 1 / (1 + Energy consumption volatility), where energy consumption volatility = historical energy consumption standard deviation / historical energy consumption mean. The lower the volatility, the higher the score, indicating more stable energy consumption. The historical energy consumption mean is the average energy consumption of the candidate equipment during the testing period, and the historical energy consumption standard deviation is determined by both the historical energy consumption mean and historical energy consumption. Cost-effectiveness score = 1 / total annual cost, where total annual cost = (initial purchase cost / estimated service life) + annual energy cost + annual maintenance cost. The annual energy cost is determined based on predicted energy consumption and unit energy cost. Initial purchase cost, annual maintenance cost, and other costs are recorded in the relevant data of the candidate equipment. The overall score is calculated as follows: M * Energy efficiency score + N * Cost-effectiveness score + Q * Reliability score. Here, M is the weight of the energy efficiency score (i.e., the first weight), N is the weight of the cost-effectiveness score (i.e., the third weight), and Q is the weight of the reliability score (i.e., the second weight). The sum of the first, second, and third weights is 1, which is set according to the actual scenario. For example, M = 0.4, N = 0.32, and Q = 0.28 can be set.

[0055] Optionally, managing the energy consumption of the target scenario based on the comprehensive score and instruction information includes: receiving instruction information, wherein the instruction information is used to instruct the reconfiguration of candidate devices in the target scenario; generating an optimal configuration scheme based on the energy consumption constraints, multiple candidate devices, and multiple comprehensive scores recorded in the instruction information, and implementing the optimal configuration scheme on the target scenario, wherein, when the optimal configuration scheme is implemented on the target scenario, the energy consumption generated by the target scenario satisfies the energy consumption constraints, and the optimal configuration scheme includes at least one candidate device.

[0056] The solution provided in this application embodiment can also manage the energy consumption of a target scenario based on a comprehensive score and specific indication information. Specifically, the indication information is received by the system from administrators or automatic triggers. This information indicates whether equipment in the target scenario (such as a data center, server room, etc.) needs to be reconfigured, and the purpose of the reconfiguration, such as reducing energy consumption, improving computing efficiency, or meeting specific energy consumption constraints. When managing the energy consumption of a target scenario according to the indication information, the energy consumption constraints of the target scenario are parsed from the indication information. The energy consumption constraint can be an energy consumption threshold or a percentage reduction in energy consumption. The energy consumption constraint is used to guide the equipment reconfiguration process, ensuring that the final configuration scheme can meet or exceed the set energy consumption constraints. For example, after determining the comprehensive score of each candidate device, a set of candidate devices that meet the energy consumption constraints is automatically selected based on the comprehensive score. Then, an optimal configuration scheme is generated using an optimization algorithm (such as a genetic algorithm, simulated annealing, etc.). The optimal configuration scheme not only considers the comprehensive score of individual devices, but also the complementarity between devices and the overall energy consumption efficiency, ensuring that the overall cost-effectiveness of the equipment configuration scheme is maximized while meeting the energy consumption constraints. In this embodiment, the optimal configuration scheme for the target scenario can be implemented by issuing configuration instructions to candidate devices in the target scenario. These configuration instructions include instructions to adjust switch states, instructions to change operating modes, etc. After implementing the optimal configuration scheme, the energy consumption of the target scenario is continuously monitored. Through real-time data collection and analysis, the actual effect of the configuration scheme is evaluated to determine whether it meets or exceeds energy consumption constraints. If energy consumption exceeds constraints or the predicted model deviates significantly from actual energy consumption, the system will automatically adjust device configurations or optimize model parameters to further improve the efficiency and accuracy of energy consumption management.

[0057] Through the above steps, it is possible to combine time series prediction models and deep learning models to further analyze complex data relationships based on the time patterns of energy consumption data, thereby shortening the time required for energy consumption prediction and improving the accuracy of energy consumption prediction. In addition, energy consumption management based on specific user instructions ensures that energy consumption is minimized and resources are optimized while meeting business needs.

[0058] Figure 3 This is a structural diagram of an energy management device provided according to an embodiment of this application, such as... Figure 3As shown, the energy management device includes: an acquisition module 30, used to acquire relevant data of candidate devices in the target scenario, wherein the relevant data includes: energy consumption data generated by the candidate devices during the detection period, the target scenario includes: a computer room, a data center, and the candidate devices include: communication equipment and basic equipment related to the operating environment of the communication equipment; a processing module 32, used to extract features from the relevant data to obtain multi-dimensional feature vectors, wherein the multi-dimensional feature vectors are used to describe multi-dimensional features related to the energy consumption of the candidate devices; a prediction module 34, used to process and analyze the multi-dimensional feature vectors using an energy consumption prediction model to obtain the energy consumption prediction value output by the energy consumption prediction model, wherein the energy consumption prediction value is used to indicate the energy consumption generated by the candidate devices during future periods, and the energy consumption prediction model includes: a time series prediction model and a deep learning model with the output of the time series prediction model as input; and a device management module 36, used to determine the comprehensive score of each candidate device based on the energy consumption prediction value and historical energy consumption value, and to manage the energy consumption of the target scenario based on the comprehensive score and indication information.

[0059] The energy management device may also include a visualization module, which contains an interactive visualization interface that supports real-time updates and interactive operations. It mainly displays information such as candidate device data tables, device energy efficiency ratio comparison bar charts, and model training loss curves.

[0060] For example, the interface can display a data table titled "Candidate Equipment List," using Dash's table tools to show equipment data, including equipment name, model, rated power, and PUE value. Pagination is supported, and users can interactively view data for different devices. An energy efficiency ratio bar chart can also be displayed, titled "Equipment Energy Efficiency Ratio Comparison," with the X-axis representing the model and the Y-axis representing the energy efficiency ratio (PUE value). Different colors are used to distinguish equipment types, facilitating comparisons of the energy efficiency ratios of different devices. A model training graph can also be created within the visualization interface, titled "Model Learning Progress," with the X-axis representing the number of training epochs. Two curves can be displayed on the graph, representing training loss and validation loss respectively. A slider component can display the loss curves at different training epochs, facilitating observation of the model's training process. Alternatively, create an interactive line chart that displays the energy consumption trends of different devices over time, titled "Device Energy Consumption Trend Chart." The X-axis represents time (hours, days, months, etc.), and the Y-axis represents energy consumption values ​​(selected energy consumption metrics: power, electricity, etc.). Different devices are represented by different colors and legends. Dash's dropdown menu component and date picker component implement device selection and time range selection functions, allowing users to select one or more devices and customize the time range (such as one week, one month, or one year) to view their energy consumption trends.

[0061] It should be noted that, Figure 3 Preferred embodiments of the shown examples can be found in [reference needed]. Figure 2 The relevant descriptions of the embodiments shown will not be repeated here.

[0062] This application also provides a non-volatile storage medium storing a computer program, wherein the above-mentioned energy management method is executed by running the computer program in the device where the non-volatile storage medium is located.

[0063] The aforementioned non-volatile storage medium is used to store programs that perform the following functions: acquiring relevant data of candidate devices in the target scenario, wherein the relevant data includes: energy consumption data generated by the candidate devices during the testing period, the target scenario includes: computer rooms and data centers, and the candidate devices include: communication equipment and basic equipment related to the operating environment of the communication equipment; extracting features from the relevant data to obtain multi-dimensional feature vectors, wherein the multi-dimensional feature vectors are used to describe multi-dimensional features related to the energy consumption of the candidate devices; processing and analyzing the multi-dimensional feature vectors using an energy consumption prediction model to obtain the energy consumption prediction value output by the energy consumption prediction model, wherein the energy consumption prediction value is used to indicate the energy consumption generated by the candidate devices during future periods, and the energy consumption prediction model includes: a time series prediction model and a deep learning model that takes the output of the time series prediction model as input; determining a comprehensive score for each candidate device based on the energy consumption prediction value and historical energy consumption value, and managing the energy consumption of the target scenario based on the comprehensive score and indication information.

[0064] This application also provides an electronic device, including a memory and a processor. The memory stores a computer program, and the processor is configured to execute the above-mentioned energy management method through the computer program.

[0065] The processor in the aforementioned electronic device is used to run a program that performs the following functions: acquiring relevant data of candidate devices in the target scenario, wherein the relevant data includes: energy consumption data generated by the candidate devices during the detection period, the target scenario includes: computer room, data center, and candidate devices include: communication equipment and basic equipment related to the operating environment of the communication equipment; extracting features from the relevant data to obtain multi-dimensional feature vectors, wherein the multi-dimensional feature vectors are used to describe multi-dimensional features related to the energy consumption of the candidate devices; processing and analyzing the multi-dimensional feature vectors using an energy consumption prediction model to obtain the energy consumption prediction value output by the energy consumption prediction model, wherein the energy consumption prediction value is used to indicate the energy consumption generated by the candidate devices during future periods, and the energy consumption prediction model includes: a time series prediction model and a deep learning model that takes the output of the time series prediction model as input; determining a comprehensive score for each candidate device based on the energy consumption prediction value and historical energy consumption value, and managing the energy consumption of the target scenario based on the comprehensive score and indication information.

[0066] This application also provides a computer program product, including computer instructions, which, when executed by a processor, implement the steps of the above-described energy management method.

[0067] It should be noted that each module in the above-mentioned energy management device can be a program module (for example, a set of program instructions to implement a certain function) or a hardware module. For the latter, it can be manifested in the following forms, but is not limited to them: each of the above modules is manifested as a processor, or the functions of each of the above modules are implemented by a processor.

[0068] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0069] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0070] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0071] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0072] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0073] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to related technologies, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0074] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. An energy consumption management method, characterized in that, include: Acquire relevant data of candidate devices in the target scenario, wherein the relevant data includes: energy consumption data generated by the candidate devices during the testing period, the target scenario includes: computer room, data center, and the candidate devices include: communication equipment and basic equipment related to the operating environment of the communication equipment; Feature extraction is performed on the relevant data to obtain a multi-dimensional feature vector, wherein the multi-dimensional feature vector is used to describe the multi-dimensional features related to the energy consumption of the candidate device; The multi-dimensional feature vector is processed and analyzed using an energy consumption prediction model to obtain the energy consumption prediction value output by the energy consumption prediction model. The energy consumption prediction value is used to indicate the energy consumption generated by the candidate device during operation in a future period. The energy consumption prediction model includes: a time series prediction model and a deep learning model that takes the output of the time series prediction model as input. A comprehensive score is determined for each candidate device based on the predicted energy consumption value and historical energy consumption value, and the energy consumption of the target scenario is managed based on the comprehensive score and indication information.

2. The method according to claim 1, characterized in that, The relevant data also includes: basic equipment data for describing the rated information of the device, and environmental data for describing the operating environment of the candidate device; Feature extraction is performed on the relevant data to obtain a multi-dimensional feature vector, including: Feature extraction is performed on different types of related data from different feature extraction dimensions to obtain multiple independent feature vectors, including: a device feature vector obtained by feature extraction of the device basic data, an environmental feature vector obtained by feature extraction of the environmental data, and an energy consumption feature vector obtained by feature extraction of the energy consumption data. Multiple independent feature vectors are concatenated to form the multi-dimensional feature vector, wherein the number of columns in the multi-dimensional feature vector is greater than the number of columns in any single independent feature vector.

3. The method according to claim 1, characterized in that, Before processing and analyzing the multi-dimensional feature vector using an energy consumption prediction model, the method includes: Identify the target elements related to energy consumption in the multi-dimensional feature vector, and combine multiple target columns to which multiple target elements belong into a target feature vector; The non-numeric elements in the target feature vector are encoded into numerical form, and the numeric elements in the target feature vector are normalized.

4. The method according to claim 1, characterized in that, The multi-dimensional feature vector is processed and analyzed using an energy consumption prediction model to obtain the energy consumption prediction value output by the energy consumption prediction model, including: The time series prediction model in the energy consumption prediction model is used to process and analyze the multi-dimensional feature vector to output energy consumption change features, wherein the energy consumption change features include: the period of energy consumption change and the energy consumption change trend in each period. The energy consumption change characteristics are processed using the deep learning model in the energy consumption prediction model, and the predicted energy consumption value is output.

5. The method according to claim 1, characterized in that, The energy consumption prediction model is trained using the following method: The historical time series stored in the distributed storage system and the historical energy consumption series corresponding to the historical time series are determined as the first type of training data, and the time series prediction model is trained using the first type of training data. The time series is used to record the historical operating time of the candidate device, and the historical energy consumption series is used to record the historical energy consumption generated by the candidate device at the historical operating time. If the training of the time series prediction model is completed, the first type of training data and the output of the time series prediction model for the first type of training data are determined as the second type of training data. The deep learning model is iteratively trained using the second type of training data. During the iterative training process, the training progress is determined based on the change in the loss function and the preset number of iterations. The training process includes: continuing training and stopping training.

6. The method according to claim 1, characterized in that, A comprehensive score is determined for each candidate device based on the predicted energy consumption and historical energy consumption values, including: For each candidate device, an energy efficiency score is determined based on the rated energy consumption of the candidate device and the predicted energy consumption value; a reliability score is determined based on the predicted energy consumption value and the historical energy consumption value; and a cost-effectiveness score is determined based on the predicted energy consumption value, unit energy consumption cost, and other preset costs. The comprehensive score is determined based on the energy efficiency score, the first weight of the energy efficiency score, the reliability score, the second weight of the reliability score, the cost-effectiveness score, and the third weight of the cost-effectiveness score.

7. The method according to claim 1, characterized in that, Managing the energy consumption of the target scenario based on the comprehensive score and indication information includes: Receive the instruction information, wherein the instruction information is used to instruct the candidate devices in the target scenario to be reconfigured; An optimal configuration scheme is generated based on the energy consumption constraints recorded in the instruction information, multiple candidate devices, and multiple comprehensive scores, and the optimal configuration scheme is implemented in the target scenario. When the optimal configuration scheme is implemented in the target scenario, the energy consumption generated by the target scenario satisfies the energy consumption constraints, and the optimal configuration scheme includes at least one candidate device.

8. An energy management device, characterized in that, include: The acquisition module is used to acquire relevant data of candidate devices in the target scenario. The relevant data includes: energy consumption data generated by the candidate devices during the detection period. The target scenario includes: computer room and data center. The candidate devices include: communication equipment and basic equipment related to the operating environment of the communication equipment. The processing module is used to extract features from the relevant data to obtain a multi-dimensional feature vector, wherein the multi-dimensional feature vector is used to describe multi-dimensional features related to the energy consumption of the candidate device. The prediction module is used to process and analyze the multi-dimensional feature vector using an energy consumption prediction model to obtain the energy consumption prediction value output by the energy consumption prediction model. The energy consumption prediction value is used to indicate the energy consumption generated by the candidate device during operation in a future period. The energy consumption prediction model includes a time series prediction model and a deep learning model that takes the output of the time series prediction model as input. The device management module is used to determine a comprehensive score for each candidate device based on the predicted energy consumption value and historical energy consumption value, and to manage the energy consumption of the target scenario based on the comprehensive score and indication information.

9. A non-volatile storage medium, characterized in that, The non-volatile storage medium stores a computer program, wherein the device containing the non-volatile storage medium executes the energy management method according to any one of claims 1 to 7 by running the computer program.

10. An electronic device comprising a memory and a processor, characterized in that, The memory stores a computer program, and the processor is configured to execute the energy management method according to any one of claims 1 to 7 through the computer program.

11. A computer program product comprising computer instructions, characterized in that, When the computer instructions are executed by the processor, they implement the steps of the energy management method according to any one of claims 1 to 7.

Images