Energy management method and energy management system

Abstract

The application relates to the technical field of energy management, and provides an energy management method and an energy management system. The method comprises the following steps: acquiring distribution design parameter information of a target park, performing 3D visualization based on the distribution design parameter information, and generating a park three-dimensional space model; marking equipment on the park three-dimensional space model to obtain N energy devices, and performing computing power network division and deployment on the N energy devices as a key point set, thereby building M distributed energy networks; calling an energy analysis network warehouse based on the M distributed energy networks to perform energy data collection and analysis processing, thereby obtaining M distributed energy utilization parameter sets; performing regulation and control strategy analysis and simulation optimization by using a digital twin technology, thereby determining energy regulation and control strategy parameters; and performing energy device response management and closed-loop control based on the strategy parameters, so that the technical problem of low energy utilization rate caused by uneven distribution of energy devices and low energy regulation and control efficiency in traditional park energy management is solved.

Description

Technical Field

[0001] This application relates to the field of energy management technology, specifically to an energy management method and an energy management system. Background Technology

[0002] In modern industrial parks and facilities, energy management has become a key factor in optimizing production efficiency and reducing operating costs. Traditional energy management models often rely on manual experience or partial automation systems, making it difficult to address issues such as the widespread distribution and heterogeneity of energy equipment and dynamic demand fluctuations. Current technologies typically employ a centralized management architecture for park energy networks, leading to uneven allocation of computing resources, equipment response delays, and an inability to optimize energy dispatch strategies in real time. Furthermore, the lack of deep integration between the physical spatial layout of energy equipment and its energy consumption characteristics makes it difficult to accurately match control strategies to actual scenario demands, resulting in widespread energy waste and hindering the improvement of energy efficiency and the achievement of carbon neutrality goals in industrial parks. Summary of the Invention

[0003] This application provides an energy management method and an energy management system, aiming to solve the technical problems of low energy utilization caused by uneven distribution of energy equipment and low energy regulation efficiency in traditional park energy management. It realizes the technical effect of using digital twin technology to simulate and optimize energy regulation strategies, improve the response efficiency of energy equipment and energy utilization efficiency, and ensure the intelligent level of energy management.

[0004] In view of the above problems, this application provides an energy management method and an energy management system.

[0005] The first aspect disclosed in this application provides an energy management method, the method comprising: acquiring distribution design parameter information of a target park; performing 3D visualization based on the distribution design parameter information to generate a three-dimensional spatial model of the park; identifying and marking energy devices in the three-dimensional spatial model of the park to obtain N energy devices, and using the N energy devices as a set of key points for computing power network partitioning and deployment to build M distributed energy networks, wherein N≥M; building an energy analysis network warehouse; based on the M distributed energy networks, calling the energy analysis network warehouse to collect and analyze energy data to obtain M sets of distributed energy utilization parameters; using digital twin technology to analyze and simulate optimization of the M sets of distributed energy utilization parameters based on the three-dimensional spatial model of the park to determine energy control strategy parameters; performing energy device response management in the target park based on the energy control strategy parameters to obtain energy device feedback parameters, and performing closed-loop control of energy devices through the energy device feedback parameters.

[0006] Another aspect of this application discloses an energy management system, the system comprising: a 3D visualization module: acquiring distribution design parameter information of a target park, performing 3D visualization based on the distribution design parameter information, and generating a three-dimensional spatial model of the park; a computing power network partitioning and deployment module: identifying and marking energy devices in the three-dimensional spatial model of the park to obtain N energy devices, and partitioning and deploying the N energy devices as a set of key points to build M distributed energy networks, where N≥M; a data acquisition and processing module: building an energy analysis network warehouse, and collecting and analyzing energy data based on the M distributed energy networks by calling the energy analysis network warehouse to obtain M sets of distributed energy utilization parameters; a strategy analysis and optimization module: using digital twin technology to analyze and simulate optimization of the control strategies of the M sets of distributed energy utilization parameters based on the three-dimensional spatial model of the park, and determining energy control strategy parameters; and a closed-loop control module: managing the response of energy devices in the target park based on the energy control strategy parameters, obtaining energy device feedback parameters, and performing closed-loop control of energy devices through the energy device feedback parameters.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] The aforementioned energy management method first acquires basic data on the distribution design parameters of the target park, including building layout, energy equipment distribution, and pipeline routing. Based on this data, a three-dimensional visualization model is constructed to create a complete three-dimensional spatial model of the park, providing an intuitive and visual foundation for energy management. Subsequently, energy equipment is identified and marked in the three-dimensional spatial model, identifying N energy devices within the park, such as power transformers, air conditioning systems, lighting equipment, and heating equipment. These energy devices are considered key nodes. By analyzing their geographical location, energy consumption characteristics, and data interaction needs, a computing network is divided and deployed, thereby building M distributed energy networks to achieve distributed management and scheduling control of the energy devices. To achieve efficient data analysis and decision support, an energy analysis network repository is further built. This repository can retrieve real-time data from the M distributed energy networks for further processing. The system collects, processes, and analyzes energy data to generate M sets of distributed energy utilization parameters, covering key indicators such as equipment operating status, energy consumption data, and load fluctuations. Based on this, digital twin technology is used to deeply integrate the actual energy operation scenario of the park with a 3D spatial model. The system then analyzes and simulates control strategies based on the M sets of energy utilization parameters. Through multiple rounds of simulation optimization, the optimal energy control strategy parameters are found to achieve refined control and optimal allocation of energy use. Finally, based on the optimized energy control strategy parameters, real-time response management is implemented for energy equipment within the park, dynamically adjusting equipment operating status and obtaining feedback parameters. This forms a closed-loop control system of data-strategy-feedback. This closed-loop control not only promptly corrects deviations in energy use but also continuously optimizes energy management strategies, ultimately achieving efficient, intelligent, green, and sustainable energy management in the park.

[0009] The above description is merely an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, specific embodiments of this application are given below. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is a flowchart illustrating an energy management method in one embodiment.

[0012] Figure 2This is a diagram of an energy management system architecture in one embodiment.

[0013] Figure labeling: 3D visualization module 11, computing power network partitioning and deployment module 12, data acquisition and processing module 13, strategy analysis and optimization module 14, closed-loop control module 15. Detailed Implementation

[0014] This application provides an energy management method and an energy management system to solve the technical problem of low energy utilization caused by uneven distribution of energy equipment and low energy regulation efficiency in traditional park energy management.

[0015] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0016] It should be noted that the terms “comprising” and “having”, and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to such process, method, product, or device.

[0017] Example 1, as Figure 1 As shown, this application provides an energy management method, the method comprising:

[0018] Obtain the distribution design parameters of the target park, perform 3D visualization based on the distribution design parameters, and generate a three-dimensional spatial model of the park.

[0019] In this embodiment, during the energy management process, it is first necessary to obtain the distribution design parameter information of the target park. This information includes key data such as the building structure layout, installation location of energy equipment, pipeline distribution, road orientation, green areas, and energy transmission paths within the park. These parameters comprehensively reflect the spatial structure and distribution of energy facilities in various areas of the park, serving as the foundational data for subsequent energy management and optimization. Subsequently, based on this distribution design parameter information, 3D modeling software (such as Revit, SketchUp, 3ds...) is applied. Software such as Max and Blender converts the data into a visualized 3D spatial model, constructing the basic structural model of the park, including buildings, roads, green spaces, and pipelines. Then, energy equipment models are precisely placed within the 3D space of the basic structural model, labeled with equipment names, numbers, and operating statuses to ensure consistency between the model and the actual distribution. Textures, materials, and lighting effects are added to the model to enhance its realism, facilitating subsequent visualization and interactive operations. Next, the basic structural model with added energy equipment is interfaced with the Energy Management System (EMS) to display real-time energy flow, equipment status, and energy consumption data, resulting in the final 3D spatial model of the park. Through this 3D visualization, managers can intuitively view the specific locations of various energy devices within the park, the connections between devices, and the energy flow paths. This eliminates the need for complex 2D drawings or cumbersome equipment data tables, allowing managers to monitor and schedule energy in an intuitive 3D scene, significantly improving management efficiency and decision-making accuracy.

[0020] Energy devices are identified and marked on the three-dimensional spatial model of the park to obtain N energy devices. The N energy devices are then used as a set of key points for the computing power network to be divided and deployed to build M distributed energy networks, where N≥M.

[0021] In one embodiment, based on the equipment names marked in the 3D spatial model of the park, all energy equipment models arranged internally are identified and uniformly labeled, resulting in N energy equipment, such as power transformers, distribution cabinets, air conditioning systems, lighting equipment, heating devices, photovoltaic equipment, etc. Subsequently, these N energy equipment are used as a set of key points, and the computational resource requirements of each key equipment are analyzed, including data acquisition frequency, processing complexity, and real-time requirements, to determine the computational resources needed for each equipment. Then, based on the park's geographical layout and equipment distribution density, a computing power network partitioning scheme is designed, thereby dividing the set of key points into M distributed energy networks. This ensures efficient and coordinated data processing and energy regulation within each network, where N ≥ M, meaning that one distributed energy network can manage multiple energy equipment simultaneously. This process not only improves the management efficiency of energy equipment in the park but also lays a solid foundation for intelligent control of energy utilization.

[0022] Furthermore, this application provides the method for constructing M distributed energy networks, including:

[0023] The process involves: acquiring information on computing power network partitioning factors, including the park's geographical layout, energy equipment distribution, and equipment data processing requirements; prioritizing each partitioning factor to obtain a reference sequence for computing power network partitioning factors; using the N energy devices as a set of key points, and evaluating and planning the computing power network for the set of key points based on the reference sequence for computing power network partitioning factors to obtain M energy computing power networks; and connecting to the computing power resource scheduling center to configure and deploy resources for the M energy computing power networks to build M distributed energy networks.

[0024] Preferably, when building M distributed energy networks, the geographical layout data of the park is first obtained, including the park's area, building distribution, road layout, and functional positioning of different areas (such as office area, production area, and storage area). This data helps determine the spatial distribution of energy equipment and its distance, thereby rationally allocating computing resources to reduce network latency and data transmission distance. Equipment distribution data for all energy equipment (such as transformers, air conditioners, lighting equipment, and heating devices) within the park is then obtained. Different types of equipment have different processing needs and importance, and the distribution of equipment directly affects the planning of the computing network. Data processing requirements data for each energy device are also obtained, including data acquisition frequency and processing complexity (such as real-time requirements, computational load, and data storage requirements). For example, some devices require real-time monitoring and rapid response, while other devices have lower processing needs. This data helps quantify the allocation and configuration of computing resources. All these acquired data will collectively form the computing network partitioning factor information, used for subsequent computing... Network assessment and planning; subsequently, based on business needs, an impact weight is assigned to each factor. Typically, equipment data processing requirements are given a higher impact weight because they directly affect the allocation and deployment of computing resources. After obtaining the impact weight of each factor, the impact weights are sorted in descending order to determine the computing network partitioning factor reference sequence. For example, equipment data processing requirements may be ranked first, followed by the distribution characteristics of energy equipment, and finally factors such as the geographical layout of the park. Then, N energy devices are taken as a set of key points, and computing network assessment and planning are carried out on the set of key points according to the computing network partitioning factor reference sequence. Specifically, because equipment data processing requirements are ranked first in the computing network partitioning factor reference sequence, the expected acquisition frequency, real-time requirements, computational load, data storage requirements, and other data processing requirement parameters of each key point are first obtained, and then these parameters are normalized (using the maximum and minimum value normalization method) to obtain data processing requirement parameters under the same dimension.Subsequently, the processed data processing requirement parameters are weighted according to the weight of each parameter (determined based on business needs and expert decisions) to obtain the quantified data processing requirement value for each key point. By comparing the quantified data processing requirement value of each key point with the preset high computing power requirement value, key points whose quantified data processing requirement value is greater than or equal to the preset high computing power requirement value are selected. These key points are directly divided into independent computing power networks (first distributed energy network, second distributed energy network, etc.) to ensure performance requirements. The remaining key points will enter the next step of selection, that is, divided according to the distribution of energy equipment ranked second. In this process, starting from any undivided key point, a first partition set is created, and other key points are traversed. Using Euclidean distance, the distance between the key point and each key point in the first partition set is calculated using the geographical coordinates of the key point. If all distances are less than or equal to the preset maximum distance, the key point is added to the first partition set; otherwise, a second partition set is created with the key point. The above process is repeated until all key points have been compared, thus obtaining multiple partition sets. Afterwards, based on the park location... The system rationally assesses each partition set. If a partition set contains key points from different functional areas, these key points are split into new partition sets to reduce cross-regional communication costs. This process is repeated until each partition set has been assessed. The key points in each assessed partition set form an energy computing network. All energy computing networks are then aggregated to obtain M energy computing networks. Each network is responsible for processing a certain number of key points, ensuring efficient resource utilization and real-time equipment response. These M distributed energy computing networks are then connected to a computing resource scheduling center. The scheduling center coordinates and allocates resources for each computing network, ensuring optimal resource allocation for the entire park's energy management system. Under the guidance of the scheduling center, resources are allocated to each energy computing network based on the network partitioning factors, resulting in M ​​energy network computing resource allocations. These energy network computing resource allocations are then combined with sensor monitoring network parameters to build M distributed energy networks. This achieves efficient and intelligent park energy management, improves energy utilization efficiency, and lays the foundation for the sustainable development of the park's energy management system.

[0025] Furthermore, this application provides the method for constructing M distributed energy networks, including:

[0026] Based on the computing power network partitioning factor information, the M energy computing power networks are identified to obtain M energy network factor parameter sets; the computing power resource scheduling center is connected to analyze the resource allocation of the M energy network factor parameter sets to obtain the computing power allocation resources of the M energy networks; according to the energy equipment monitoring requirements, the sensor parameters of the M energy computing power networks are analyzed to obtain M sensor monitoring network parameters; based on the computing power allocation resources of the M energy networks and the M sensor monitoring network parameters, the resources of the M energy computing power networks are configured and deployed to build the M distributed energy networks.

[0027] Optionally, after obtaining M energy computing power networks, the device data processing requirements are extracted from the computing power network segmentation factor information, including the quantified value of the data processing requirements for each key point. By summing the quantified values ​​of the data processing requirements for the key points involved in each energy computing power network, the total data processing requirements for each energy computing power network are obtained. This total data processing requirements are then used to identify the energy computing power networks, resulting in a set of M energy network factor parameters. Subsequently, a communication connection is established with the computing power resource scheduling center, and the set of M energy network factor parameters is transmitted to the computing power resource scheduling center. The computing power resource scheduling center will then process the M... Resource allocation is analyzed using energy network factor parameter sets. This process iterates through M energy network factor parameter sets and multiplies the obtained parameters by the unit energy network computing power resources stored internally in the computing power resource scheduling center. This yields M energy network computing power allocation resources. Here, unit energy network computing power resources refer to standardized unit resources defined within the computing power resource scheduling center, representing the basic unit of resource allocation required by each energy network. This unit is used to quantify and allocate the computing power needs of different energy networks. For example, unit energy network computing power resources could be 1 CPU core + 4GB. The system consists of a memory module, 10MB / s bandwidth, and 100GB storage. Then, based on the types of devices and monitoring needs in each energy computing network, the required sensor types and parameters are determined. For example, according to the monitoring requirements of the devices, a suitable sampling frequency is set; for devices with high accuracy requirements, high-precision sensors are selected to ensure the accuracy of data acquisition. The communication protocols and bandwidth requirements of the sensors are determined to ensure that sensor data can be smoothly transmitted to the energy management system. By summarizing the sensor parameters of each energy computing network, M sensor monitoring network parameters are obtained, providing detailed parameter information for subsequent sensor deployment. After obtaining the energy network computing power allocation resources and sensor monitoring network parameters, resources (such as computing nodes and storage devices) are deployed to each energy computing network to ensure that all devices in the network can obtain computing resources as needed. Appropriate sensor devices are also installed according to the sensor parameters and connected to the corresponding energy network to ensure real-time monitoring. This completes the construction of the network topology, resulting in the resource configuration and deployment of M distributed energy networks. This ensures that each network has the necessary computing power resources and sensor support, ultimately achieving efficient energy management and improving the efficiency and reliability of the overall energy management system.

[0028] An energy analysis network repository is established. Based on the M distributed energy networks, the energy analysis network repository is called to collect and analyze energy data, resulting in M ​​sets of distributed energy utilization parameters.

[0029] In one embodiment, an energy analysis network repository is established, which is built based on relevant analysis networks of energy equipment, such as energy consumption analysis networks and energy efficiency analysis networks. Based on multiple distributed energy networks (i.e., M distributed energy networks) within the park, each containing a different set of energy equipment, the energy data from these devices can be used to call the corresponding analysis networks in the energy analysis network repository for data collection and analysis. In this way, energy data can not only be accurately captured but also processed by advanced algorithms to generate corresponding energy utilization parameter sets. Specifically, each distributed energy network will determine the appropriate energy utilization parameter set based on the type of equipment it contains and... During operation, the system sends requests to the energy analysis network repository to retrieve analysis networks related to these devices. For example, if a distributed energy network includes multiple transformers, air conditioning systems, and lighting equipment, it will call upon the energy analysis networks of the corresponding devices in the repository to perform energy consumption analysis and energy efficiency prediction. The energy analysis networks in the repository perform a series of calculations based on the specific parameters collected for each device, ultimately obtaining various parameters related to energy utilization, including energy consumption and efficiency data. These parameters are fed back to each distributed energy network in the form of an energy utilization parameter set, enabling the park's energy management system to monitor the real-time operation of each network and device. Through continuous data collection, analysis, and processing, energy allocation can be optimized, resource utilization efficiency improved, and energy dispatch strategies adjusted through the data feedback mechanism to ensure that energy use within the park operates under optimal conditions.

[0030] Furthermore, this application provides the aforementioned method for building an energy analysis network repository, including:

[0031] Based on the N energy devices, an energy utilization characteristic task list is constructed, which includes energy consumption, energy efficiency, and energy cost analysis tasks for each energy device. Historical data sets of energy device utilization are collected and acquired. These historical data sets are then segmented and labeled based on the N energy devices to obtain N energy device utilization sample sets. Transfer analysis training is performed on the N energy device utilization sample sets based on the energy utilization characteristic task list to obtain N energy device analysis networks. Finally, the N energy device analysis networks are encoded, labeled, and stored in a data warehouse to obtain the energy analysis network warehouse.

[0032] Preferably, the energy utilization characteristic task list serves as the foundation for system analysis of each energy device. It includes various tasks and indicators related to device operation, such as energy consumption, energy efficiency, and energy cost analysis tasks. Specifically, the energy consumption analysis task involves the energy consumption of each device within a specific time period; the energy efficiency analysis task focuses on the device's energy efficiency ratio (EER), which is the ratio between the energy consumed and the work performed or the benefits generated; and the energy cost analysis task assesses the operating cost of each device. By defining these tasks, a detailed task list can be created for each device, facilitating subsequent data collection, analysis, and optimization. Subsequently, a large amount of historical data is obtained from the historical database. This data includes device runtime, energy consumption, load conditions, and efficiency performance. This data constitutes the energy device's historical utilization dataset, which is then used... The historical data of energy device utilization for N energy devices is distributed as follows: First, using the identification information of each energy device, historical utilization data for each device is extracted from the historical data dataset, forming N energy device utilization sample sets. Then, based on the constructed list of energy utilization feature tasks and the collected historical data, a transfer learning algorithm is used to train and generate an energy device analysis network for each device. During this process, the N energy device utilization sample sets are input into the matching analysis networks for analysis and training, resulting in N energy device analysis networks. These energy device analysis networks can predict the energy efficiency and energy consumption of specific devices, providing decision support for the energy management system. Finally, the generated N energy device analysis networks are coded and identified (to which device they belong) and stored in an energy analysis network repository for subsequent retrieval and management.

[0033] Furthermore, this application provides the aforementioned analysis network for obtaining N energy devices, including:

[0034] Based on the energy utilization characteristic task list, a source domain energy characteristic network set is selected, which includes an energy consumption analysis network, an energy efficiency analysis network, and an energy cost analysis network. The energy consumption analysis network, energy efficiency analysis network, and energy cost analysis network are then connected and fused in parallel to obtain the source domain energy analysis network. Model parameters are extracted from the source domain energy analysis network to determine the model architecture parameters. Based on the model architecture parameters, transfer learning training and optimization are performed on the N energy device utilization sample sets to obtain the N energy device analysis networks.

[0035] Optionally, based on the task requirements in the energy utilization feature task list, existing or similar relatively accurate energy feature analysis models (built based on neural networks, decision trees, etc.) are searched from the model library, and the searched models are stored as a source domain energy feature network set. The models in this source domain energy feature network set correspond one-to-one with the tasks in the energy utilization feature task list, such as an energy consumption analysis network corresponding to the energy consumption task, an energy efficiency analysis network corresponding to the energy efficiency task, and an energy cost analysis network corresponding to the energy cost task. Subsequently, the energy consumption analysis network, energy efficiency analysis network, and energy cost analysis network are connected in parallel to form a multi-task joint network, namely the source domain energy analysis network. This network can simultaneously analyze energy consumption, efficiency, and cost, providing comprehensive data support for subsequent equipment optimization. After the source domain energy analysis network is fused, the weights and biases of each sub-network are extracted from the fused network. Model architecture parameters, such as the number of layers and neurons per layer, represent the model's complexity and learning ability, serving as the foundation for transfer learning and allowing for optimization during subsequent training. Then, using the extracted architecture parameters, transfer learning training is performed on a sample set of energy utilization data from the devices. Transfer learning leverages knowledge from the source domain (existing energy feature analysis models) to help quickly customize and optimize models for new devices or application scenarios. By using the architecture and parameters of the source domain energy analysis network, transfer learning training is performed on energy device utilization samples. That is, the energy device utilization samples are input into the source domain energy analysis network, and the weights are gradually adjusted and the network optimized through forward propagation, loss calculation, backpropagation, and parameter optimization to adapt it to new device characteristics and requirements, resulting in N energy device analysis networks. These analysis networks will help devices optimize energy utilization in real time, providing accurate decision support for subsequent energy management systems.

[0036] Using digital twin technology, the control strategy of the M distributed energy utilization parameter sets is analyzed and optimized through simulation based on the three-dimensional spatial model of the park, and the energy control strategy parameters are determined.

[0037] In one embodiment, an energy management database is constructed to collect and organize historical data, equipment performance, load conditions, and energy consumption information of various distributed energy networks within the park. This data provides a foundation for subsequent control strategies. Subsequently, based on the energy management database, an energy control status prediction network is established through supervised training. This network, by learning from historical data, can predict the operating status of various energy systems within the park under different conditions, helping to identify potential energy waste, inefficient areas, or equipment. Then, the prediction network is embedded into a three-dimensional spatial model of the park, and digital twin technology is used for simulation to generate an energy status digital twin model, thereby mapping the park's energy utilization in real time. This virtual environment, simulating the impact of different control strategies on the energy system, provides an operational space for energy dispatch. Based on this, control strategies are analyzed for M sets of distributed energy utilization parameters. Various possible control schemes are analyzed according to energy management objectives (such as improving energy efficiency, reducing energy consumption, and lowering costs). Through the analysis of these schemes, a set of energy control strategy parameter solutions is obtained. Finally, based on the generated digital twin model of energy state, these control strategy parameter solutions are simulated and optimized. That is, by simulating the implementation effects of different strategies, their performance in actual operation is evaluated to find the optimal energy control strategy. This process helps the park formulate the best energy management plan, thereby optimizing energy utilization, improving efficiency, and reducing costs. Through the above process, the park can not only monitor energy usage in real time but also dynamically adjust and optimize according to different needs and operating conditions, ensuring the efficient and stable operation of the energy management system.

[0038] Furthermore, this application provides the parameters for determining the energy regulation strategy, including:

[0039] An energy management database is constructed, and a deep learning network is used to predict and supervise the training of the database to obtain an energy regulation state prediction network. This network is then embedded into a three-dimensional spatial model of the park using digital twin technology to generate a digital twin model of the energy state. The M distributed energy utilization parameter sets are analyzed according to the energy regulation demand objectives to obtain a set of energy regulation strategy parameters. Finally, the energy regulation strategy parameters are optimized through simulation based on the digital twin model of the energy state.

[0040] Optionally, historical energy data within the park can be obtained through data mining, including energy consumption, efficiency, operating time, load status, and environmental factors of various equipment, to construct an energy management database. This database provides necessary input for subsequent deep learning training. Subsequently, a deep learning network (such as a multilayer perceptron or recurrent neural network) is used to train the constructed energy management database. The training method is the same as described above, involving forward propagation, loss calculation, backpropagation, and parameter optimization. The goal of the training process is to enable the network to accurately predict energy regulation states under different conditions, such as energy consumption, efficiency changes, and costs. By continuously adjusting the model parameters, the network gradually learns how to extract valuable features from historical data, forming an energy regulation state prediction network. This network can assess the operating status of various energy systems within the park in real time, providing decision support for energy management. Once the energy regulation state prediction network is obtained, the next step is to use digital twin technology to embed it into the park's three-dimensional spatial model. The park's three-dimensional spatial model is not merely the physical layout of the equipment. This also includes the operating characteristics and energy consumption patterns of the equipment. By combining the energy regulation state prediction network with a three-dimensional spatial model, a digital twin model of energy status is formed. This virtual model can dynamically simulate the operation of various energy systems within the park, making energy scheduling and management more intuitive and accurate. It also provides a virtual environment for subsequent testing of regulation strategies. Next, based on the energy regulation demand target, the park's M distributed energy utilization parameter sets are analyzed. That is, based on historical data, multiple energy regulation strategies that can achieve the M distributed energy utilization parameter sets are obtained (the specific number can be determined according to actual needs), thus obtaining the energy regulation strategy parameter solution set. Then, based on the previously obtained energy status digital twin model, the energy regulation strategy parameter solution set is simulated. The simulation evaluates the effect of different strategies in actual operation, simulates the impact of each strategy on energy consumption, energy efficiency, energy cost, etc., and optimizes by comparing the effects of different strategies. The purpose of this process is to find the optimal regulation strategy so that the park's energy management system can achieve maximum energy utilization efficiency while maintaining system stability.

[0041] Furthermore, this application provides the parameters for determining the energy regulation strategy, including:

[0042] Based on the energy regulation demand target, an energy regulation effect fitness function is constructed; using the energy regulation effect fitness function, the energy state digital twin model is used to simulate and evaluate each parameter solution in the energy regulation strategy parameter solution set, obtaining the regulation effect fitness of multiple parameter solutions; based on the regulation effect fitness of the multiple parameter solutions, the energy regulation strategy parameter solution set is expanded by cross-mutation to obtain the regulation strategy parameter solution space; global optimization is performed in the regulation strategy parameter solution space to determine the energy regulation strategy parameters.

[0043] Optionally, based on the park's energy regulation and control demand targets, an energy regulation and control effectiveness fitness function is constructed. This function is a quantitative evaluation of the effectiveness of different regulation and control strategies, reflecting the advantages and disadvantages of the strategies in meeting various demand targets. The fitness function is generally obtained by weighted summation. For example, the energy regulation and control effectiveness fitness function is as follows: Where F represents the adaptability of energy regulation effect, , , These are the weighting coefficients for each objective, reflecting the importance of each objective; E is the energy consumption parameter. It is the target energy consumption demand. It is an energy efficiency parameter. It is a target efficiency requirement. It is an energy cost parameter. The target cost requirement is determined by the following steps: First, the solutions from the energy regulation strategy parameter set are input into the energy state digital twin model for simulation. The simulation results are then passed to the internal energy regulation state prediction network to generate energy regulation state prediction results for each parameter solution, including energy consumption, energy efficiency, and energy cost. These prediction results are then input into the energy regulation effect fitness function to calculate the fitness of multiple parameter solutions. Next, crossover and mutation are performed based on the fitness values ​​to generate more combinations of regulation strategies. Crossover combines two preferred strategy solutions to generate new solutions; for example, combining the parameters of two strategies to form a new regulation strategy parameter solution. Mutation randomly changes some parameters in the existing solutions to explore new solution spaces and avoid getting trapped in local optima. After crossover and mutation, new regulation strategy parameter solutions are generated, and these solutions together constitute the regulation strategy parameter solution space. These new solutions will contain more possibilities, making it easier to find the optimal solution. After obtaining the expanded solution space for the control strategy parameters, the next step is to perform global optimization to select the best control strategy. In this process, global optimization algorithms (such as particle swarm optimization, genetic algorithms, etc.) are used to search within the expanded solution space to find the strategy solution with the highest fitness. Taking particle swarm optimization as an example, the particle swarm optimization algorithm represents each potential solution in the solution space as a particle, and each particle has its own position and velocity. In each iteration, the particle adjusts its velocity and position based on its own historical best position and global best position, thereby expanding the solution space. The optimal solution is found within the swarm. Specifically, during each generation update, the particles evaluate the quality of the current solution by calculating the fitness function of energy regulation effect, and use information exchange to guide the entire particle swarm to converge toward the optimal solution. The position of each particle represents a solution of regulation strategy parameters. The fitness function of energy regulation effect calculates the quality of each solution according to the target requirements (such as energy efficiency, cost, etc.). Through multiple generations of iteration, the particle swarm will gradually find the most suitable regulation strategy. The final optimized regulation strategy parameters will be used as the energy regulation strategy parameters to determine the specific energy dispatch scheme within the park, ensuring that the energy management system maintains efficient and stable operation while achieving the optimization goal.

[0044] Based on the energy regulation strategy parameters, energy equipment response management is performed on the target park to obtain energy equipment feedback parameters, and closed-loop control of energy equipment is performed through the energy equipment feedback parameters.

[0045] In one embodiment, adjustment commands are issued to each energy device within the park based on determined energy control strategy parameters to ensure they operate according to the new control strategy. As the devices begin executing these commands, feedback parameters are generated. These parameters reflect the actual operating status of the devices under the new strategy, including energy consumption feedback, efficiency feedback, and cost feedback. These feedback parameters are collected in real time, forming energy device feedback parameters, and transmitted back to the energy management system for further monitoring and adjustment. This achieves closed-loop control of the energy devices. The goal of closed-loop control is to continuously optimize and adjust the operating status of the devices based on this feedback data, ensuring optimal energy use within the park and promptly correcting any potential deviations or anomalies. Through this feedback control mechanism, the system can dynamically respond to changes in energy demand within the park and continuously self-regulate, thereby achieving more precise and efficient energy management.

[0046] In summary, the embodiments of this application have at least the following technical effects:

[0047] This application embodiment obtains the distribution design parameter information of the target park, performs 3D visualization based on the distribution design parameter information, and generates a three-dimensional spatial model of the park; identifies and marks energy equipment in the three-dimensional spatial model of the park to obtain N energy equipment, and uses the N energy equipment as a key point set for computing power network partitioning and deployment to build M distributed energy networks, where N≥M; builds an energy analysis network warehouse, and uses the energy analysis network warehouse to collect and analyze energy data based on the M distributed energy networks to obtain M distributed energy utilization parameter sets; uses digital twin technology to analyze and simulate optimization of the control strategy based on the three-dimensional spatial model of the park to determine energy control strategy parameters; performs energy equipment response management in the target park based on the energy control strategy parameters to obtain energy equipment feedback parameters, and performs closed-loop control of energy equipment through the energy equipment feedback parameters. These technologies collectively address the technical problems of low energy utilization caused by uneven distribution of energy equipment and low energy regulation efficiency in traditional park energy management. They enable the use of digital twin technology to simulate and optimize energy regulation strategies, improve the response efficiency of energy equipment and the efficiency of energy utilization, and ensure the level of intelligent energy management.

[0048] Example 2, based on the same inventive concept as the energy management method in the foregoing examples, such as... Figure 2As shown, this application provides an energy management system, the system comprising: a 3D visualization module 11: acquiring the distribution design parameter information of the target park, performing 3D visualization based on the distribution design parameter information, and generating a three-dimensional spatial model of the park; a computing power network partitioning and deployment module 12: identifying and marking energy devices in the three-dimensional spatial model of the park to obtain N energy devices, and using the N energy devices as a set of key points for computing power network partitioning and deployment to build M distributed energy networks, where N≥M; a data acquisition and processing module 13: building an energy analysis network warehouse, and using the energy analysis network warehouse to collect and analyze energy data based on the M distributed energy networks to obtain M sets of distributed energy utilization parameters; a strategy analysis and optimization module 14: using digital twin technology to analyze and simulate the optimization of the control strategy based on the three-dimensional spatial model of the park for the M sets of distributed energy utilization parameters, and determining the energy control strategy parameters; and a closed-loop control module 15: performing energy device response management in the target park based on the energy control strategy parameters, obtaining energy device feedback parameters, and performing closed-loop control of energy devices through the energy device feedback parameters.

[0049] Furthermore, the computing power network partitioning and deployment module 12 is also used to perform the following method:

[0050] The process involves: acquiring information on computing power network partitioning factors, including the park's geographical layout, energy equipment distribution, and equipment data processing requirements; prioritizing each partitioning factor to obtain a reference sequence for computing power network partitioning factors; using the N energy devices as a set of key points, and evaluating and planning the computing power network for the set of key points based on the reference sequence for computing power network partitioning factors to obtain M energy computing power networks; and connecting to the computing power resource scheduling center to configure and deploy resources for the M energy computing power networks to build M distributed energy networks.

[0051] Furthermore, the computing power network partitioning and deployment module 12 is also used to perform the following method:

[0052] Based on the computing power network partitioning factor information, the M energy computing power networks are identified to obtain M energy network factor parameter sets; the computing power resource scheduling center is connected to analyze the resource allocation of the M energy network factor parameter sets to obtain the computing power allocation resources of the M energy networks; according to the energy equipment monitoring requirements, the sensor parameters of the M energy computing power networks are analyzed to obtain M sensor monitoring network parameters; based on the computing power allocation resources of the M energy networks and the M sensor monitoring network parameters, the resources of the M energy computing power networks are configured and deployed to build the M distributed energy networks.

[0053] Furthermore, the data acquisition and processing module 13 is also used to perform the following methods:

[0054] Based on the N energy devices, an energy utilization characteristic task list is constructed, which includes energy consumption, energy efficiency, and energy cost analysis tasks for each energy device. Historical data sets of energy device utilization are collected and acquired. These historical data sets are then segmented and labeled based on the N energy devices to obtain N energy device utilization sample sets. Transfer analysis training is performed on the N energy device utilization sample sets based on the energy utilization characteristic task list to obtain N energy device analysis networks. Finally, the N energy device analysis networks are encoded, labeled, and stored in a data warehouse to obtain the energy analysis network warehouse.

[0055] Furthermore, the data acquisition and processing module 13 is also used to perform the following methods:

[0056] Based on the energy utilization characteristic task list, a source domain energy characteristic network set is selected, which includes an energy consumption analysis network, an energy efficiency analysis network, and an energy cost analysis network. The energy consumption analysis network, energy efficiency analysis network, and energy cost analysis network are then connected and fused in parallel to obtain the source domain energy analysis network. Model parameters are extracted from the source domain energy analysis network to determine the model architecture parameters. Based on the model architecture parameters, transfer learning training and optimization are performed on the N energy device utilization sample sets to obtain the N energy device analysis networks.

[0057] Furthermore, the strategy parsing and optimization module 14 is also used to perform the following method:

[0058] An energy management database is constructed, and a deep learning network is used to predict and supervise the training of the database to obtain an energy regulation state prediction network. This network is then embedded into a three-dimensional spatial model of the park using digital twin technology to generate a digital twin model of the energy state. The M distributed energy utilization parameter sets are analyzed according to the energy regulation demand objectives to obtain a set of energy regulation strategy parameters. Finally, the energy regulation strategy parameters are optimized through simulation based on the digital twin model of the energy state.

[0059] Furthermore, the strategy parsing and optimization module 14 is also used to perform the following method:

[0060] Based on the energy regulation demand target, an energy regulation effect fitness function is constructed; using the energy regulation effect fitness function, the energy state digital twin model is used to simulate and evaluate each parameter solution in the energy regulation strategy parameter solution set, obtaining the regulation effect fitness of multiple parameter solutions; based on the regulation effect fitness of the multiple parameter solutions, the energy regulation strategy parameter solution set is expanded by cross-mutation to obtain the regulation strategy parameter solution space; global optimization is performed in the regulation strategy parameter solution space to determine the energy regulation strategy parameters.

[0061] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0062] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0063] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. An energy management method, characterized in that, The method includes: Obtain the distribution design parameter information of the target park, perform 3D visualization based on the distribution design parameter information, and generate a three-dimensional spatial model of the park; Energy devices are identified and marked on the three-dimensional spatial model of the park to obtain N energy devices. The N energy devices are then used as a set of key points for computing power network partitioning and deployment to build M distributed energy networks, where N≥M. An energy analysis network repository is established, and energy data is collected and analyzed based on the M distributed energy networks to obtain M sets of distributed energy utilization parameters. Using digital twin technology, the control strategy of the M distributed energy utilization parameter sets is analyzed and simulated based on the three-dimensional spatial model of the park to determine the energy control strategy parameters. Based on the energy regulation strategy parameters, energy equipment response management is performed on the target park to obtain energy equipment feedback parameters, and closed-loop control of energy equipment is performed through the energy equipment feedback parameters. The establishment of the energy analysis network repository includes: Based on the N energy devices, construct an energy utilization characteristic task list, which includes energy consumption, energy efficiency and energy cost analysis tasks for each energy device. Collect and obtain a historical dataset of energy equipment utilization, and then perform flow identification on the historical dataset of energy equipment utilization based on the N energy equipment to obtain a sample set of energy equipment utilization. Based on the energy utilization feature task list, transfer analysis training is performed on the N energy equipment utilization sample sets to obtain N energy equipment analysis networks. The N energy equipment analysis networks are coded, identified, and stored in a data warehouse to obtain the energy analysis network warehouse.

2. The energy management method as described in claim 1, characterized in that, The construction of M distributed energy networks includes: Obtain information on computing power network partitioning factors, including the geographical layout of the park, the distribution of energy equipment, and the data processing requirements of the equipment. The priority of each partitioning factor in the computing power network partitioning factor information is sorted to obtain a reference sequence of computing power network partitioning factors; The N energy devices are taken as a set of key points. Based on the reference sequence of computing power network partitioning factors, the set of key points is evaluated and planned to obtain M energy computing power networks. The computing power resource scheduling center is connected to configure and deploy resources for the M energy computing power networks respectively, and builds M distributed energy networks.

3. The energy management method as described in claim 2, characterized in that, The construction of M distributed energy networks includes: Based on the computing power network partitioning factor information, the M energy computing power networks are identified to obtain M energy network factor parameter sets; The computing power resource scheduling center is connected to analyze the resource allocation of the M energy network factor parameter sets to obtain the computing power allocation resources of the M energy networks. According to the energy equipment monitoring requirements, the sensor parameters of the M energy computing networks are analyzed to obtain the M sensor monitoring network parameters. Based on the computing power allocation resources of the M energy networks and the parameters of the M sensor monitoring networks, the resources of the M energy computing power networks are configured and deployed to build the M distributed energy networks.

4. The energy management method as described in claim 1, characterized in that, The obtained N energy device analysis network includes: Based on the energy utilization characteristic task list, a source domain energy characteristic network set is selected, which includes an energy consumption analysis network, an energy efficiency analysis network, and an energy cost analysis network. The energy consumption analysis network, energy efficiency analysis network, and energy cost analysis network are connected and merged in parallel to obtain the source domain energy analysis network. Model parameters are extracted from the source domain energy analysis network to determine the model architecture parameters; Based on the model architecture parameters, the N energy devices are trained and optimized using transfer learning with a sample set to obtain the analysis network for the N energy devices.

5. The energy management method as described in claim 1, characterized in that, The determination of energy regulation strategy parameters includes: An energy management database is constructed by mining and a deep learning network is used to perform predictive supervised training on the energy management database to obtain an energy regulation state prediction network. The energy regulation state prediction network is embedded into the three-dimensional spatial model of the park using digital twin technology to perform twin simulation and generate a digital twin model of energy state. According to the energy regulation demand target, the regulation strategy is analyzed for the M distributed energy utilization parameter sets to obtain the energy regulation strategy parameter solution set; Based on the aforementioned digital twin model of energy state, the energy regulation strategy parameter set is simulated and optimized to determine the energy regulation strategy parameters.

6. The energy management method as described in claim 5, characterized in that, The determination of energy regulation strategy parameters includes: Based on the energy regulation demand target, construct an energy regulation effect fitness function; The energy regulation effect fitness function is used to simulate and evaluate the parameter solutions in the energy regulation strategy parameter solution set based on the energy state digital twin model, and the regulation effect fitness of multiple parameter solutions is obtained. Based on the fitness of the multiple parameter solutions to control effects, the energy control strategy parameter solution set is cross-mutated and expanded to obtain the control strategy parameter solution space. Global optimization is performed within the solution space of the regulation strategy parameters to determine the energy regulation strategy parameters.

7. An energy management method, characterized in that, The steps for implementing an energy management method according to any one of claims 1 to 6 include: 3D visualization module: acquires the distribution design parameter information of the target park, performs 3D visualization based on the distribution design parameter information, and generates a three-dimensional spatial model of the park; Computing power network partitioning and deployment module: energy equipment is identified and marked in the three-dimensional spatial model of the park to obtain N energy equipment, and the N energy equipment are used as a set of key points for computing power network partitioning and deployment to build M distributed energy networks, where N≥M; Data acquisition and processing module: Builds an energy analysis network warehouse, and uses the energy analysis network warehouse to collect and analyze energy data based on the M distributed energy networks, thereby obtaining M sets of distributed energy utilization parameters; Strategy Analysis and Optimization Module: Utilizing digital twin technology, based on the three-dimensional spatial model of the park, the module analyzes and simulates the regulation strategies of the M distributed energy utilization parameter sets to determine the energy regulation strategy parameters. Closed-loop control module: Based on the energy regulation strategy parameters, it performs energy equipment response management on the target park, obtains energy equipment feedback parameters, and performs closed-loop control of energy equipment through the energy equipment feedback parameters.

Images