Energy management methods, systems, media, and products based on collaboration between AI and energy gateways

Abstract

An energy management method, system, medium, and product for collaborative AI and energy gateways relate to the field of computer systems based on specific computing models. The method includes: obtaining historical operating data of each energy gateway in an energy gateway group, generating an energy management model, and generating an energy equipment management strategy; assigning the newly added energy gateway to the corresponding gateway type group based on its gateway feature data; generating simulated operating data for the newly added energy gateway based on the configuration information of the gateway type group, and incrementally training the energy management model to generate a simulated energy model, obtain simulated scheduling instructions and simulated parameter configurations to adjust the energy equipment management strategy; obtaining actual operating data, determining simulation deviation parameters, and adjusting the simulated energy model to obtain a target energy model; and optimizing the energy equipment management strategy based on the target energy model. Implementing this application can improve the scheduling and control effect of the newly added energy gateway during the initial operation phase.

Description

Technical Field

[0001] The present application relates to the field of computer systems based on specific computing models, and in particular to an energy management method, system, medium and product that collaborates with AI and energy gateways. Background Art

[0002] With the rapid development of the Energy Internet, energy gateways, as a crucial bridge between energy devices and energy management systems, are playing an increasingly important role in the overall energy management system. Energy gateways must not only collect and control data from energy devices but also support intelligent device scheduling and optimize operating parameters to ensure efficient operation of the energy system. In scenarios where large-scale energy devices are connected, leveraging the management capabilities of energy gateways to improve the overall efficiency of energy systems has become a key industry concern.

[0003] In related technologies, energy management systems primarily manage energy gateways based on pre-set rules. System administrators pre-define a series of fixed scheduling rules and parameter configurations based on experience, and the energy gateways execute corresponding control instructions according to these pre-set rules. Furthermore, the energy management system collects operational data from the energy gateways and performs simple data analysis and alarm processing based on pre-set thresholds, thus achieving basic monitoring and management of energy equipment.

[0004] However, when a new energy gateway is connected to the system, due to the lack of operational data accumulation, the system can only adopt general management strategies for scheduling and control, and the scheduling and control effect is poor. Summary of the Invention

[0005] This application provides an energy management method, system, medium and product that collaborates with AI and energy gateways, which are used to improve the scheduling and control effect of new energy gateways in the initial operation stage.

[0006] In a first aspect, the present application provides an energy management method for collaboration between AI and energy gateways, which is applied to an energy management system, the method comprising: obtaining historical operation data of each energy gateway in an energy gateway group, and generating an energy management model through a machine learning algorithm; based on the energy management model, generating an energy equipment management strategy for the energy gateway group, including energy scheduling instructions and operation parameter configurations; determining gateway feature data based on access information of a newly added energy gateway to the energy gateway group, and assigning the newly added energy gateway to a corresponding gateway type group based on the gateway feature data; generating simulated operation data of the newly added energy gateway based on the configuration information of the gateway type group; incrementally training the energy management model based on the simulated operation data to generate a simulated energy model; adjusting the energy scheduling instructions and operation parameter configurations based on the simulated energy model to obtain simulated scheduling instructions and simulated parameter configurations to adjust the energy equipment management strategy; obtaining actual operation data of the newly added energy gateway, determining simulation deviation parameters of the simulated operation data, and adjusting the simulated energy model based on the simulation deviation parameters to obtain a target energy model; based on the target energy model, correcting the simulated scheduling instructions and simulation parameter configurations to actual scheduling instructions and actual parameter configurations to optimize the energy equipment management strategy.

[0007] In the above embodiment, the energy management system generates an energy management model by acquiring historical operating data, and classifies and simulates the operation based on the characteristics of the new gateway, thereby achieving rapid adaptation and precise control of the new energy gateway; through incremental training of simulated operation data and deviation correction of actual operation data, the energy management strategy can quickly adapt to the operating characteristics of the new gateway, effectively improving the scheduling and control effect of the new gateway in the initial operation stage.

[0008] In combination with some embodiments of the first aspect, in some embodiments, the step of generating an energy equipment management strategy for an energy gateway group including energy scheduling instructions and operating parameter configurations based on an energy management model specifically includes: obtaining multidimensional operating data of each energy gateway in the energy gateway group; the multidimensional operating data includes power generation data, power consumption data, equipment efficiency data, meteorological environment data and load distribution data; inputting the multidimensional operating data into the energy management model to obtain timing prediction data of the energy gateway group; the timing prediction data includes a power generation timing curve, a power demand timing curve and a load fluctuation timing curve; calculating the load balancing coefficient of the energy gateway group based on the timing prediction data; the load balancing coefficient is used to characterize the degree of matching between the power generation capacity and power demand of each energy gateway; based on the load balancing coefficient, generating energy scheduling instructions and operating parameter configuration data for the energy gateway group to obtain an energy equipment management strategy.

[0009] In the above embodiment, the energy management system realizes the refined management of the energy gateway group by collecting multi-dimensional operation data and performing time series prediction, and generating a scheduling strategy in combination with the load balancing coefficient. It can accurately grasp the power generation and consumption characteristics of each gateway, reasonably distribute the load, and improve the overall operation efficiency.

[0010] In combination with some embodiments of the first aspect, in some embodiments, after the step of generating energy scheduling instructions and operating parameter configuration data of the energy gateway group according to the load balancing coefficient and obtaining the energy equipment management strategy, the method also includes: obtaining historical fault data of each energy gateway; the historical fault data includes fault type, fault time, fault duration and fault environment parameters; performing correlation analysis on the historical fault data based on the energy management model to obtain equipment fault correlation feature data, and generating a fault warning threshold based on the equipment fault correlation feature data; the equipment fault correlation feature data includes fault occurrence frequency, fault impact range and fault propagation path; the fault warning threshold includes operating parameter threshold, environmental parameter threshold and equipment status threshold; when it is detected that the operating data of any energy gateway exceeds the corresponding fault warning threshold, generating linkage control data of the energy gateway group, and sending the linkage control data to the corresponding energy gateway; the linkage control data includes load transfer instructions, standby startup instructions and fault isolation instructions.

[0011] In the above embodiment, the energy management system establishes an early warning mechanism by analyzing historical fault data and adjusts the operation strategy in time before a fault occurs, thereby realizing active prevention and control of the system, quickly responding to abnormal situations, and ensuring stable operation of the system.

[0012] In combination with some embodiments of the first aspect, in some embodiments, after the step of respectively correcting the simulated scheduling instructions and simulated parameter configurations to actual scheduling instructions and actual parameter configurations based on the target energy model to optimize the energy equipment management strategy, the method also includes: receiving a policy adjustment instruction input by the user through the management terminal; the policy adjustment instruction includes optimization target configuration, energy usage preference, operating time period setting and cost control parameters; the optimization target configuration includes energy saving priority, cost priority, comfort priority and peak-valley balance priority; constructing a user personalized model based on the policy adjustment instruction, and integrating the user personalized model with the target energy model for training to obtain a personalized energy model; optimizing the actual scheduling instructions and actual parameter configurations based on the personalized energy model to obtain personalized scheduling instructions and personalized parameter configurations; monitoring the execution result data of the personalized scheduling instructions and personalized parameter configurations in real time, and calculating the user satisfaction index based on the optimization target and execution result data set by the user; and rebuilding the personalized energy model when the user satisfaction index is lower than the preset threshold during the continuous monitoring period.

[0013] In the above embodiment, the energy management system introduces user personalized needs, and realizes dynamic optimization of energy management strategies through model fusion and satisfaction monitoring. It can flexibly adjust the operation plan according to user preferences, thereby improving user experience.

[0014] In combination with some embodiments of the first aspect, in some embodiments, after the step of generating an energy device management strategy for an energy gateway group including energy scheduling instructions and operating parameter configurations based on an energy management model, the method further includes: monitoring the online status of each energy gateway in the energy gateway group, and when it is detected that the target energy gateway is in an offline state, obtaining the historical operating data and associated gateway data of the target energy gateway; the associated gateway data includes the gateway identifier, interactive energy type, interactive energy amount and interactive timing data that have energy interaction with the target energy gateway; determining the energy supply and demand characteristics of the target energy gateway based on the historical operating data to obtain the target gateway characteristic data; calculating the impact coefficient of the target energy gateway offline on the energy gateway group based on the target gateway characteristic data and the associated gateway data; updating the parameters of the energy management model based on the impact coefficient to obtain an offline adjustment model, and regenerating the energy device management strategy based on the offline adjustment model.

[0015] In the above embodiment, the energy management system analyzes historical data and associated gateway data to evaluate the impact of gateway offline conditions, thereby achieving adaptive adjustment of the system and ensuring that the group can still operate efficiently when the gateway is abnormal.

[0016] In combination with some embodiments of the first aspect, in some embodiments, after the steps of updating the parameters of the energy management model based on the influence coefficient to obtain an offline adjustment model, and regenerating the energy equipment management strategy based on the offline adjustment model, the method also includes: obtaining load redundancy data of each energy gateway in the energy gateway group; the load redundancy data is used to characterize the backup energy capacity of each energy gateway; based on the load redundancy data, selecting a target backup gateway from the energy gateway group; calculating the device startup timing and energy output curve of the target backup gateway, and generating a backup startup instruction; the backup startup instruction includes the device startup time, energy output power and output duration; sending the backup startup instruction to the target backup gateway, and obtaining the real-time operation data of the target backup gateway; adjusting the prediction parameters of the offline adjustment model according to the real-time operation data, so that the deviation between the prediction result of the offline adjustment model and the real-time operation data is less than a preset error range.

[0017] In the above embodiment, the energy management system selects a backup gateway through load redundancy analysis, thereby achieving fault-tolerant recovery of the system and ensuring the reliability of the backup solution through real-time monitoring and model adjustment.

[0018] In combination with some embodiments of the first aspect, in some embodiments, before the step of generating simulated operation data of the newly added energy gateway based on the configuration information of the gateway type group, the method also includes: extracting the device parameters and operation characteristics of the newly added energy gateway to obtain gateway verification data; the gateway verification data includes device model, specification parameters, power characteristics, communication protocol type and control mode; constructing a verification rule set based on the gateway verification data, and verifying the type of the newly added energy gateway based on the verification rule set to obtain a verification result; when the verification result shows that there is a deviation, recalculating the type characteristic value of the newly added energy gateway based on the gateway verification data.

[0019] In the above embodiment, the energy management system ensures the accuracy of gateway classification through device parameter verification, provides a reliable basis for subsequent simulation operations, and improves the stability of system operation.

[0020] In a second aspect, an embodiment of the present application provides an energy management system, which includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the energy management system to execute the method described in the first aspect and any possible implementation method of the first aspect.

[0021] In a third aspect, an embodiment of the present application provides a computer program product comprising instructions, which, when executed on an energy management system, enables the energy management system to execute the method described in the first aspect and any possible implementation of the first aspect.

[0022] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium comprising instructions. When the instructions are executed on an energy management system, the energy management system executes the method described in the first aspect and any possible implementation of the first aspect.

[0023] It is understood that the energy management system provided in the second aspect, the computer program product provided in the third aspect, and the computer storage medium provided in the fourth aspect are all used to execute the methods provided in the embodiments of the present application. Therefore, the beneficial effects that can be achieved can be referenced to the beneficial effects of the corresponding methods and will not be repeated here.

[0024] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0025] 1. By adopting an energy management model training based on historical data and a feature classification mechanism for newly added gateways, combined with incremental training based on simulated operation data and deviation correction based on actual operation data, the system can quickly identify the operating characteristics of newly added gateways and generate an adaptive management strategy. This effectively solves the problem of poor scheduling effect of newly added gateways in the initial stage in the existing technology, and thus realizes intelligent collaborative management of energy gateway groups, significantly improving the access efficiency and initial operation effect of newly added gateways.

[0026] 2. Due to the adoption of multi-dimensional operation data collection and time series prediction analysis, and the method of optimizing scheduling strategies based on load balancing coefficients, the system can fully grasp the operating status of each gateway and accurately predict changes in energy supply and demand, effectively solving the problems of extensive energy scheduling and unbalanced load distribution in existing technologies, thereby realizing the refined management of gateway groups and load optimization configuration, greatly improving the overall operating efficiency of the system.

[0027] 3. Due to the adoption of the mechanism of gateway offline detection, historical data analysis and associated gateway impact assessment, and dynamic adjustment of the model based on the impact coefficient, the system can promptly identify abnormal situations and respond, effectively solving the problem of a single gateway abnormality affecting the overall performance of the system in the existing technology, thereby realizing the adaptive adjustment and stable operation of the system, and ensuring that the energy gateway group can still maintain efficient collaboration in the event of local abnormalities. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 This is a flowchart of the energy management method using AI and energy gateway collaboration in an embodiment of the present application;

[0029] Figure 2 This is another flowchart of the energy management method using AI and energy gateway collaboration in an embodiment of the present application;

[0030] Figure 3 This is a schematic diagram of the structure of a physical device of the energy management system in an embodiment of the present application. DETAILED DESCRIPTION

[0031] The terms used in the following examples of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the specification of the present application, the singular expressions "a", "an", "above", "the", and "this" are intended to include plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in the present application refers to any or all possible combinations of one or more of the listed items.

[0032] In the following, the terms "first" and "second" are used for descriptive purposes only and should not be understood to imply or suggest relative importance or implicitly indicate the number of the technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of this application, unless otherwise specified, "plurality" means two or more.

[0033] For ease of understanding, the application scenarios of the embodiments of the present application are introduced below.

[0034] An industrial park boasts multiple distributed energy facilities, including rooftop photovoltaic systems, wind turbines, energy storage power stations, and smart power loads. As the park expands, the number of energy facilities continues to grow, significantly increasing system complexity. Especially when new equipment is added, the system needs to quickly identify the device type and adapt the control strategy. For example, the park recently added a hybrid energy storage system consisting of electrochemical and flywheel energy storage, whose charging and discharging characteristics differ significantly from those of traditional energy storage devices. Furthermore, different businesses have different energy requirements. For example, manufacturing companies prioritize energy supply reliability, while R&D companies prioritize energy comfort. When a system failure occurs, rapid fault diagnosis and emergency response are required to prevent the failure from spreading and causing widespread power outages. These complex scenarios place higher demands on the intelligence of energy management systems.

[0035] In related technologies, simple parameter matching and threshold monitoring methods can be used to manage and diagnose energy devices. For example, newly added devices can be matched using preset device type templates, device failures can be identified using fixed parameter thresholds, and energy devices can be managed using a unified scheduling strategy. This approach is simple to implement, but it cannot adapt to the complex and ever-changing needs of energy system management. The following describes a scenario using the energy management method based on related technologies, which combines AI with an energy gateway.

[0036] Traditional energy management systems use fixed management models to address these issues. For example, the energy management system used by a chemical park could only match newly added equipment using preset device type templates, failing to flexibly adapt to the characteristics of new equipment. When a new energy storage inverter was connected, the system was unable to accurately identify its type due to its unique grid-connection characteristics and control mode, resulting in a control strategy mismatch. Regarding fault handling, the system could only detect faults based on simple threshold judgments and was unable to analyze the correlation between faults. For example, when a photovoltaic inverter failure caused grid fluctuations, the system failed to promptly identify the fault propagation path, ultimately leading to a cascading failure of multiple electrical devices. Furthermore, the system's unified energy scheduling strategy failed to optimize and adjust to the individual needs of different users, resulting in inefficient energy utilization.

[0037] The energy management method using AI and energy gateway collaboration in the embodiments of this application achieves intelligent identification, fault prediction, and personalized management of energy equipment by building a multi-dimensional device feature analysis model, a fault association warning model, and a user personalized model. This method not only improves the adaptability and reliability of the system, but also meets the personalized needs of different users. The following describes scenarios using the energy management method using AI and energy gateway collaboration in this application.

[0038] The energy management system using the solution of this application has shown significant advantages in actual application in a certain high-tech industrial park. When a new intelligent microgrid system is added to the park, the system accurately identifies the functional modules such as photovoltaic power generation, energy storage and load control contained therein through multi-dimensional feature analysis, and automatically configures the corresponding control strategy. In terms of fault warning, the system establishes a fault correlation model between devices by analyzing historical fault data. For example, when an abnormal output voltage fluctuation of a certain energy storage inverter is detected, the system immediately analyzes the range of equipment that may be affected, adjusts the operating parameters of the relevant equipment in advance, and successfully avoids the spread of the fault. At the same time, the system can dynamically optimize the energy scheduling strategy according to the needs of different users. For example, for enterprises that focus on energy conservation, the system gives priority to the use of renewable energy; for enterprises with higher requirements for power supply quality, the system gives priority to ensuring power quality indicators.

[0039] It can be seen that the energy management method that adopts the collaboration of AI and energy gateway in the embodiment of this application can not only realize intelligent equipment management, but also effectively solve the problems existing in traditional methods such as inaccurate equipment identification, untimely fault warning, and difficulty in meeting personalized needs, thereby realizing efficient and reliable operation of the energy system.

[0040] For ease of understanding, the following describes the process of the method provided by this implementation in combination with the above scenario. Figure 1 , which is a flow chart of the energy management method of collaboration between AI and energy gateway in an embodiment of the present application.

[0041] S101. Obtain historical operating data of each energy gateway in the energy gateway group, and generate an energy management model through a machine learning algorithm.

[0042] Among them, the energy gateway group refers to a collection of multiple energy gateways with similar functions or geographical locations; historical operation data refers to the complete data records generated by the energy gateway during its past operation, including energy production data, energy consumption data, equipment status data and environmental parameter data, etc.; machine learning algorithms are used to represent algorithms that can learn and extract characteristic patterns from historical data, including but not limited to neural networks, decision trees, etc.; energy management models refer to mathematical models used to guide the operation and scheduling of energy equipment, including sub-models such as energy supply and demand forecasting models, load balancing models and scheduling optimization models.

[0043] During the initialization phase, the energy management system needs to establish a basic energy management model for subsequent energy scheduling and management. Specifically, the energy management system first uses the data acquisition module to obtain historical operating data from all energy gateways within the group. This includes energy production data (such as photovoltaic power generation and energy storage charge and discharge), energy consumption data (such as power load curves and peak-valley distribution), device operating status data (such as equipment efficiency and fault records), and environmental parameter data (such as temperature, humidity, and light intensity). The energy management system then uses a preprocessing module to clean, standardize, and extract features from the raw data, generating a standardized training dataset. Finally, the energy management system feeds the processed data into a machine learning algorithm for training. Through multiple rounds of iterative optimization, an energy management model with predictive and decision-making capabilities is developed.

[0044] In some embodiments, the energy management model generation process can be implemented in a variety of ways. Optionally, the energy management system can employ deep learning methods, building a multi-layer neural network and combining it with a time series prediction algorithm to extract features and recognize patterns from historical data, thereby establishing a composite energy management model comprised of multiple sub-models. This method first performs time series decomposition and feature mapping on the data, then learns the time series features using a recurrent neural network, and finally outputs prediction results through a fully connected layer. Alternatively, the energy management system can employ an ensemble learning approach, combining multiple base learners (such as random forests and XGBoost) to leverage the strengths of different models and build an energy management model with strong generalization capabilities. This method first trains multiple base models of different types and then fuses the prediction results of these models through weighted voting or stacking. It is understood that other machine learning methods can also be employed to generate energy management models, and these are not limited here.

[0045] During model training, the energy management system encounters data sparsity, meaning that data from certain energy gateways is missing or of poor quality during certain time periods, impacting model training effectiveness. To address this, the energy management system employs a solution that combines data augmentation and transfer learning: first, missing data is supplemented using a time series interpolation algorithm. Then, transfer learning is performed using data features from other gateways under similar operating conditions to enhance data richness. Specifically, the system establishes a similarity calculation method based on dynamic time warping (DTW) to identify gateways with similar operating characteristics. Using their complete historical data as a reference, the system then supplements the target gateway's missing data through feature mapping. This approach significantly improves data quality and ensures the effectiveness of model training.

[0046] S102: Based on the energy management model, generate an energy device management policy for the energy gateway group, including energy scheduling instructions and operation parameter configurations.

[0047] Among them, energy dispatch instructions refer to a specific set of commands that control the operating status of energy equipment, including start-stop control instructions, power adjustment instructions, and operating mode switching instructions; operating parameter configuration represents the set values of various adjustable operating parameters of energy equipment, including output power, operating time period, control threshold, etc.; energy equipment management strategy is used to represent a complete control plan that guides the coordinated operation of energy equipment groups, including dispatching strategies, parameter configurations, optimization goals, etc.

[0048] After establishing the energy management model, the energy management system needs to generate specific management strategies to guide equipment operation. Specifically, the energy management system first acquires real-time operational data, including multi-dimensional data such as power generation, power load, energy storage status, and equipment efficiency, and inputs it into the energy management model for analysis. Based on its internal forecasting algorithm, the model generates forecast curves for power generation, power demand, and energy storage capacity changes for future time periods. The energy management system then uses a multi-objective optimization algorithm based on these forecast results, combined with load balancing requirements and economic objectives, to calculate the optimal equipment operation plan. Finally, the optimization results are converted into specific scheduling instructions and parameter configurations, forming a complete energy equipment management strategy.

[0049] In some embodiments, management strategies can be generated through a variety of methods. Optionally, the energy management system can employ a model-predictive control (MPC)-based approach to dynamically generate scheduling strategies through rolling optimization. This approach first establishes a dynamic model of the system to predict energy supply and demand for multiple time periods in the future, then solves for the optimal control sequence under constraints. Alternatively, the energy management system can employ reinforcement learning to learn optimal strategies through the interaction between the agent and the environment. This approach models the energy scheduling problem as a Markov decision process and optimizes scheduling decisions through algorithms such as deep Q-learning. It is understood that other optimization algorithms can also be employed to generate management strategies, and these are not limited here.

[0050] During the strategy generation process, energy management systems often encounter challenges in balancing multi-objective optimization, such as conflicts between economic efficiency and reliability. To address this, energy management systems employ a hierarchical optimization approach: First, at the macro level, weights for each optimization objective are determined, establishing a comprehensive evaluation function. Then, at the meso level, the optimization problem is broken down into multiple sub-problems, each solved separately. Finally, at the micro level, strategies are coordinated and conflicts resolved. For example, the system can first optimize economic indicators to obtain a basic strategy, then modify the strategy based on reliability constraints. Through iterative optimization, the system ultimately arrives at a management strategy that balances all objectives.

[0051] S103: Determine gateway characteristic data according to access information of the newly added energy gateway to the energy gateway group, and assign the newly added energy gateway to a corresponding gateway type group based on the gateway characteristic data.

[0052] Among them, access information represents the basic information of a new energy gateway when it joins the group, including device type, communication parameters, physical location, etc.; gateway feature data refers to a set of key parameters that can characterize the characteristics of the gateway, including power characteristics, control mode, communication protocol, etc.; the gateway type group is used to represent a classified set of energy gateways with similar characteristics, which is used to realize classification management and policy reuse.

[0053] When the energy management system detects a new energy gateway requesting access, it needs to perform feature analysis and type classification. Specifically, the energy management system first parses the access request from the newly added gateway to obtain basic device information and communication parameters. Then, based on pre-set feature extraction rules, the system extracts key characteristic parameters from this basic information, including the device's rated power, operating mode, control accuracy, response time, and so on. The system then performs a similarity match between these characteristic parameters and the feature templates of existing gateway type groups to calculate the best matching type group. Finally, the system assigns the newly added gateway to the corresponding type group, and inherits the basic configuration information of that group.

[0054] In some embodiments, gateway classification can be achieved through a variety of methods: Optionally, the energy management system can employ a clustering-based approach, determining the most similar type groups by calculating distances in the feature space. This approach first constructs feature vectors and then automatically classifies them using algorithms such as K-means or hierarchical clustering. Alternatively, the energy management system can employ a rule-based classification approach, using a pre-set classification rule tree to determine and categorize gateway features layer by layer. It is understood that other classification methods can also be employed to achieve gateway type attribution, and these are not limited here.

[0055] During the gateway classification process, the energy management system may encounter the problem of newly added gateways' features not adequately matching existing type groups. To address this, the energy management system employs a dynamic type group creation mechanism: when the similarity between a newly added gateway and all existing type groups falls below a threshold, the system automatically creates a new type group. Specifically, the system first calculates the feature similarity between the newly added gateway and each type group. If the maximum similarity falls below a preset threshold, the system uses the gateway's features as a template to create a new type group and initializes the default configuration information for that group. This approach ensures accurate gateway classification while maintaining the scalability of the classification system.

[0056] S104: Generate simulation operation data of the newly added energy gateway according to the configuration information of the gateway type group.

[0057] Among them, the configuration information represents the standard operating parameters and control strategy set of the gateway type group, including operating mode configuration, parameter range setting, control logic definition, etc.; the simulated operation data refers to the gateway operation data obtained through simulation calculation, which is used to predict the operating performance of the newly added gateway, including energy output curve, load response characteristics, control effect evaluation and other data.

[0058] After completing gateway classification, the energy management system needs to generate initial simulated operating data to evaluate its operational characteristics. Specifically, the energy management system first reads the standard configuration information for the corresponding type group, including parameter configuration, control strategies, and performance indicators for typical operating scenarios. Then, based on the specific parameters of the newly added gateway, the system uses a mathematical model to simulate operating conditions under different operating conditions. The system considers various influencing factors, such as load changes, environmental conditions, and operating time, to generate a complete simulated operating data set, including energy production and consumption curves, device response characteristics, and control accuracy indicators.

[0059] In some embodiments, simulated operating data can be generated in a variety of ways: Optionally, the energy management system can employ a physical model-based simulation method to simulate the dynamic characteristics of the device by establishing a detailed mathematical model. This method takes into account the physical, thermodynamic, and electrical characteristics of the device and accurately reflects the device's operating patterns. Optionally, the energy management system can also employ a data-driven simulation method to predict the operating performance of newly added gateways by analyzing historical operating data of similar gateways and combining similarity mapping. It is understood that other simulation methods can also be employed to generate operating data, and this is not limited here.

[0060] During the simulation data generation process, the energy management system encountered issues where certain characteristic parameters of newly added gateways differed significantly from the type group standard. To address this, the energy management system implemented an adaptive parameter adjustment mechanism. First, a parameter deviation correction model was established to analyze the differences between the newly added gateways and the type group standard. Simulation parameters were then dynamically adjusted based on the degree of discrepancy, ensuring that the simulation results more accurately reflected actual conditions. For example, if the response time of a newly added gateway was detected to be significantly longer than the type group average, the system would adjust the control model's time constant accordingly, ensuring that the simulation data more accurately reflected the device's characteristics.

[0061] S105 : Incrementally train the energy management model based on the simulated operation data to generate a simulated energy model.

[0062] Among them, incremental training refers to a training method that performs local updates based on the existing model without retraining the entire model; the simulated energy model refers to the new model obtained after incremental training, which includes the prediction and control models of the newly added gateway characteristics.

[0063] The energy management system needs to integrate the simulated data from the newly added gateway into the existing energy management model. Specifically, the system first preprocesses and extracts features from the simulated operating data to ensure compatibility with the existing model. Then, the system uses an incremental learning algorithm to perform a partial update training of the existing model using the simulated data. During the training process, the system automatically adjusts the model parameter weights to accurately describe the operating characteristics of the newly added gateway. Finally, the system verifies the prediction accuracy and generalization capabilities of the updated model and generates a simulated energy model that includes the newly added gateway.

[0064] In some embodiments, incremental model training can be achieved through a variety of methods: Alternatively, the energy management system can employ online learning methods to gradually update model parameters through streaming data processing, enabling real-time response to changes in the characteristics of new data. Alternatively, the energy management system can employ transfer learning methods to transfer the feature representation capabilities of existing models to new scenarios through knowledge transfer. It is understood that other incremental learning methods can also be employed to update the model, and these are not limited here.

[0065] During incremental training, the energy management system can encounter the problem of overfitting the model to simulated data. To address this, the system employs a combination of regularization and validation. First, L1 / L2 regularization terms are introduced during training to limit excessive updates to model parameters. Model performance is then evaluated through cross-validation to ensure good generalization. Furthermore, the system saves model checkpoints before incremental training, allowing for rapid rollback to the previous state if performance degradation is detected.

[0066] S106 , adjusting the energy dispatching instructions and the operating parameter configurations based on the simulated energy model to obtain simulated dispatching instructions and simulated parameter configurations to adjust the energy equipment management strategy.

[0067] Among them, the simulation scheduling instruction refers to the preliminary scheduling command generated based on the simulated energy model, which is used to guide the operation control of the newly added gateway; the simulation parameter configuration refers to the operation parameter setting that matches the simulation scheduling instruction to ensure that the scheduling instruction can be effectively executed.

[0068] Based on the updated simulated energy model, the energy management system needs to generate a scheduling strategy that adapts to the newly added gateways. Specifically, the energy management system first uses the simulated energy model to predict the system's operating status after the new gateways are added, including multiple dimensions such as energy supply and demand balance, load distribution, and network stability. The system then adjusts the existing scheduling instructions and parameter configurations based on the predicted results, generating a simulated scheduling plan that takes into account the characteristics of the newly added gateways. This process comprehensively considers multiple objectives, including economic efficiency, reliability, and environmental protection, and uses a multi-objective optimization algorithm to determine the optimal scheduling strategy.

[0069] In some embodiments, scheduling policy adjustments can be implemented in a variety of ways: Optionally, the energy management system can employ a gradual adjustment approach, initially maintaining the existing policy framework and optimizing only the scheduling instructions directly related to the newly added gateway, then gradually expanding the adjustment scope until the system as a whole reaches a new equilibrium state. Optionally, the energy management system can also employ a global reoptimization approach, taking into account the complete system model after the addition of the gateway and recalculating the optimal scheduling policy. It is understood that other optimization methods can also be employed to adjust the scheduling policy, and these are not limited here.

[0070] During the policy adjustment process, the energy management system encountered the problem of sudden increases in local network load due to the introduction of new gateways. To address this, the energy management system implemented a time-based, gradual adjustment approach. First, the scheduling cycle was divided into multiple time periods, and the load distribution ratio was gradually adjusted within each time period. Furthermore, load change rate limits were set to ensure a smooth and controllable load transfer process. In practice, the system established a timeline-based load transfer curve and used a smoothing function to control the load change rate, preventing system instability caused by overly rapid scheduling policy adjustments.

[0071] S107: Acquire actual operation data of the newly added energy gateway, determine simulation deviation parameters of the simulated operation data, and adjust the simulated energy model based on the simulation deviation parameters to obtain a target energy model.

[0072] Among them, the actual operation data refers to the real data collected by the newly added gateway during the actual operation process; the simulation deviation parameter refers to the difference characteristics between the simulated operation data and the actual operation data, including static deviation and dynamic deviation; the target energy model is used to represent the final model after correction by the actual data.

[0073] After the newly added gateway begins actual operation, the energy management system needs to calibrate the model using real data. Specifically, the energy management system first collects actual operating data from the newly added gateway, including energy production data, device status data, and control response data. The system then compares and analyzes this actual data with previously generated simulated operating data, calculating deviation parameters for various indicators. The system analyzes the time-varying characteristics of the deviations and their relevance to operating conditions to establish a comprehensive deviation characteristic model. Finally, based on this deviation characteristic model, the simulated energy model is corrected to obtain a more accurate target energy model.

[0074] In some embodiments, model correction can be achieved through a variety of methods: Optionally, the energy management system can employ an error compensation method, establishing a deviation compensator to correct the model output. This method can quickly respond to deviations in actual operation. Alternatively, the energy management system can employ an adaptive learning method, dynamically updating model parameters through online parameter estimation, enabling the model to adapt to actual operating characteristics. It is understood that other correction methods can also be employed to optimize model accuracy, and these are not limited here.

[0075] During the model correction process, the energy management system may encounter abnormal operating conditions in the actual operating data. To address this, the energy management system employs a robust correction solution: first, anomaly detection is performed on the actual operating data to identify and filter abnormal operating condition data. Then, model correction is performed based on a confidence-weighted approach, assigning different weights to data from different operating conditions. For example, the system can establish multiple operating condition sub-models and dynamically switch or merge these sub-models based on the similarity of the current operating conditions, thereby improving model adaptability.

[0076] S108. Based on the target energy model, respectively modify the simulated scheduling instructions and simulated parameter configurations to actual scheduling instructions and actual parameter configurations to optimize the energy equipment management strategy.

[0077] Among them, the actual dispatch instruction represents the final dispatch command after actual operation verification; the actual parameter configuration refers to the optimal operating parameter setting that matches the actual dispatch instruction; and the optimized energy equipment management strategy is used to represent the complete management plan after actual operation verification and optimization.

[0078] The energy management system must ultimately optimize its management strategy based on the target energy model. Specifically, it first uses the target energy model to reassess the system's operating status and predict various performance indicators. The system then compares the simulated scheduling instructions and parameter configurations with the actual operating results, analyzes the causes of any discrepancies, and identifies key areas for optimization. Based on the characteristics and constraints reflected in the actual operating data, the system then fine-tunes the scheduling instructions and parameter configurations to ensure that the optimized management strategy best meets actual operational needs.

[0079] In some embodiments, policy optimization can be achieved through a variety of methods: Optionally, the energy management system can employ an iterative optimization method based on actual performance feedback, continuously improving the policy through multiple rounds of actual operation verification. Alternatively, the energy management system can employ a targeted optimization method based on sensitivity analysis, focusing on optimizing key parameters with significant performance impact. It is understood that other optimization methods can also be employed to improve the management policy, and these are not limited here.

[0080] During the strategy optimization process, energy management systems often encounter the interdependent constraints of multiple performance indicators. To address this, the energy management system employs a hierarchical optimization approach: First, the priority of each performance indicator is determined, establishing a hierarchical optimization target system. Then, optimization is performed step by step based on priority, prioritizing lower-priority targets while maintaining higher-priority targets. For example, the system might first ensure the fundamental goal of balancing energy supply and demand, then optimize economic indicators based on this foundation, and finally consider long-term goals such as equipment lifespan. This hierarchical optimization approach achieves balanced improvements across all indicators.

[0081] The following is a supplement to the scenario of this embodiment.

[0082] During ongoing operation, this solution's energy management system has demonstrated significant optimization potential. For example, a technology park has introduced a blockchain-based energy trading mechanism based on its existing system. The system automatically identifies energy gateways with trading capabilities and optimizes energy allocation based on the trading strategies of different users. When a user has excess photovoltaic power, the system automatically identifies other users with electricity needs and develops the optimal trading plan based on real-time electricity prices and user preferences. Regarding fault prevention and control, the system continuously refines its fault warning model by learning from accumulated fault cases. For example, if the system identifies that a certain type of inverter is prone to failure under specific weather conditions, it will adjust operating parameters in advance of similar weather conditions. Furthermore, the system can continuously optimize its personalized model based on user feedback. For example, if it discovers that a company is particularly sensitive to power quality during peak production periods, it will automatically increase power quality control standards during those periods.

[0083] After combining the above scenarios, the following is a more detailed description of the process of the method provided by this implementation. Figure 2 , which is another flow chart of the energy management method of collaboration between AI and energy gateway in an embodiment of the present application.

[0084] S201. Obtain historical operating data of each energy gateway in the energy gateway group, and generate an energy management model through a machine learning algorithm.

[0085] Referring to step S101 , the energy management system constructs an energy management model.

[0086] It's important to note that the energy management system uses not only historical operating data but also external data sources such as weather data and user behavior data when training its models. Using deep learning algorithms, the system trains and optimizes various energy models. For example, it uses years of historical weather data and solar panel power generation data from a specific time period to train a solar power generation forecast model, while also training an electricity demand forecast model based on long-term user electricity usage data. This approach creates energy management models with enhanced predictive accuracy and generalizability.

[0087] S202: Acquire multi-dimensional operation data of each energy gateway in the energy gateway group.

[0088] Multidimensional operational data refers to the multi-dimensional data set generated by energy gateways during real-time operation, including power generation data (such as photovoltaic power generation and wind power generation), power consumption data (such as load power and power consumption), equipment performance data (such as equipment efficiency and conversion efficiency), meteorological and environmental data (such as temperature, humidity, and light intensity), and load distribution data (such as peak and valley distribution and power consumption patterns). An energy gateway group is a collection of multiple energy gateways with similar functions or geographical locations.

[0089] After completing the initial construction of the energy management model, the energy management system needs to obtain real-time operating data to guide scheduling decisions. Specifically, the energy management system first uses the data acquisition unit in each energy gateway to collect real-time operating data from various energy devices. For solar panels, this collects data such as light intensity, generated power, and panel temperature; for wind turbines, this collects data such as wind speed, wind direction, generator speed, and output power; for energy storage batteries, this collects parameters such as charge and discharge status, remaining capacity, battery voltage, and current; and for electrical devices, this collects real-time power, power consumption, and operating status data. This data is transmitted to the energy management system via the energy gateway's communication unit using a pre-defined uplink communication protocol (such as DLT698-45 or DLT645).

[0090] In some embodiments, the collection of multi-dimensional data can be achieved in a variety of ways: Optionally, the energy management system can adopt a layered collection method, first collecting raw data through sensors at the device layer, then performing data aggregation and preprocessing at the gateway layer, and finally performing data integration at the system layer. This method first configures the sampling frequency and collection parameters of various types of sensors, then performs data standardization and time synchronization through the gateway, and finally achieves unified management of data; Optionally, the energy management system can also adopt an event-driven collection method, performing on-demand collection based on device status changes and threshold triggers, and achieving dynamic data collection by setting collection trigger conditions and response strategies. It is understandable that other data collection methods can also be used to obtain multi-dimensional operating data, which is not limited here.

[0091] During data collection, the energy management system often encounters the problem of data synchronization being out of sync, leading to poor data relevance. To address this, the energy management system employs a timestamp-based data synchronization mechanism. First, a unified clock source is configured for all gateways to ensure consistent time bases. The collected data is then timestamped and aligned based on the timestamps. Finally, an interpolation algorithm is used to fill in any data gaps, ensuring data continuity and integrity. For example, if data at a particular time point is missing, the system performs linear interpolation based on data from nearby time points to maintain a continuous data sequence.

[0092] S203 , inputting multi-dimensional operation data into the energy management model to obtain time series prediction data of the energy gateway group.

[0093] Time-series forecasting data represents the predicted operating status of the energy system over a period of time. These include a time-series curve for power generation (reflecting future power generation trends), a time-series curve for power demand (predicting load fluctuations), and a time-series curve for load fluctuation (describing load dynamics). Energy management models are mathematical models trained using machine learning methods and used for forecasting and decision-making. These models include sub-models such as energy supply and demand forecasting models, load balancing models, and scheduling optimization models.

[0094] After acquiring multidimensional operational data, the energy management system needs to utilize energy management models for predictive analysis. Specifically, the energy management system first preprocesses the collected multidimensional data, including data cleaning, standardization, and feature extraction. The processed data is then organized into time series and fed into the corresponding submodels of the energy management model. For example, environmental data such as light intensity and temperature are fed into a photovoltaic power generation prediction model to predict future photovoltaic power generation; historical electricity usage data and current load conditions are fed into a power demand prediction model to predict future electricity demand. Finally, the system integrates the prediction results of each submodel to generate a complete time series prediction dataset.

[0095] In some embodiments, time series forecasting can be achieved through a variety of methods. Optionally, the energy management system can employ deep learning methods, using long short-term memory (LSTM) networks to process time series data features. This method first constructs a multi-layer LSTM network structure, using an attention mechanism to capture long-term data dependencies, and then performs sequence forecasting using a sliding window approach. Alternatively, the energy management system can employ an ensemble learning approach, combining the strengths of multiple basic forecasting models. This method first trains multiple different types of forecasting models (such as ARIMA and Prophet), and then synthesizes the forecast results of each model through weighted fusion. It is understood that other forecasting methods can also be used to generate time series forecast data, and this is not limited here.

[0096] During the forecasting process, energy management systems often encounter the problem of gradually decreasing forecast accuracy over time. To address this, the energy management system employs a multi-scale forecasting strategy: First, the forecast timeframe is divided into three scales: short-term (hourly), medium-term (dayly), and long-term (weekly). Different forecasting models and parameter configurations are then employed for each timescale. Finally, a progressive forecasting approach is employed, using short-term forecast results to continuously refine medium- and long-term forecasts, improving overall forecast accuracy. For example, the system updates the short-term forecast results hourly and adjusts the load forecast curve for the day accordingly to ensure the timeliness of the forecast results.

[0097] S204: Calculate the load balancing coefficient of the energy gateway group based on the time series prediction data.

[0098] The load balancing coefficient is an indicator parameter used to quantitatively assess the balance between energy supply and demand. It represents the degree of match between the power generation capacity of each energy gateway and its power demand. A coefficient closer to 1 indicates a higher degree of supply and demand matching, while a deviation from 1 indicates a supply and demand imbalance.

[0099] Energy management systems need to assess the system's supply-demand balance based on time-series forecast data. Specifically, the system first aligns the predicted power generation time series curve with the power demand time series curve and calculates the supply-demand ratio at each point in time. It then considers the regulation capabilities of energy storage devices and analyzes the impact of their charge and discharge capacities at different times on the supply-demand balance. Finally, a weighted calculation is performed to derive a load balancing coefficient that reflects the overall balance state, taking into account both temporal and spatial balance characteristics.

[0100] In some embodiments, the calculation of the load balancing coefficient can be achieved in a variety of ways: Optionally, the energy management system can adopt a hierarchical and partitioned calculation method to first calculate the local balancing coefficient of a single gateway, and then calculate the global balancing coefficient of the gateway group. This method first evaluates the supply and demand balance status within the coverage area of each gateway, and then considers the energy complementarity between gateways, and finally derives the group-level balancing index; Optionally, the energy management system can also adopt a dynamic weighting method to dynamically adjust the calculation weight according to the importance of different time periods and different types of loads. This method flexibly reflects the system operation characteristics by setting a time-varying weight matrix. It is understandable that other calculation methods can also be used to obtain the load balancing coefficient, which is not limited here.

[0101] During the calculation process, the energy management system may encounter severe fluctuations in localized load that can affect the overall load balancing assessment. To address this, the energy management system employs a robustness assessment mechanism: first, it establishes an abnormal fluctuation detection model to identify and filter out unusual load changes; then, it introduces a smoothing factor to reduce the impact of instantaneous fluctuations on the load balancing coefficient; finally, by setting multiple evaluation time windows, it integrates short-term fluctuations and long-term trends to achieve a more stable load balancing assessment result. For example, the system simultaneously calculates 5-minute, 15-minute, and 1-hour moving average load balancing coefficients for scheduling decisions at different time scales.

[0102] S205: Generate energy dispatch instructions and operating parameter configuration data for the energy gateway group based on the load balancing coefficient to obtain an energy device management strategy.

[0103] Energy dispatch instructions represent a specific set of commands that control the operating status of energy devices, including start / stop control instructions, power regulation instructions, and operating mode switching instructions. Operating parameter configuration data refers to the parameter settings required for device operation, including output power limits, operating time settings, and control thresholds. Energy device management strategies are complete control plans that guide the coordinated operation of groups of energy devices.

[0104] The energy management system needs to develop appropriate management strategies based on the load balancing coefficient. Specifically, the energy management system first determines the system's supply and demand balance based on the load balancing coefficient. If the coefficient deviates from 1, the system analyzes the cause and extent of the deviation. Based on this analysis, the system automatically generates appropriate control strategies, including adjusting the output power of power generation equipment, controlling the charging and discharging behavior of energy storage devices, and optimizing the start and stop sequences of power-consuming equipment. Finally, the system converts these strategies into specific scheduling instructions and parameter configurations, which are then distributed to each device via the energy gateway.

[0105] In some embodiments, the generation of management strategies can be achieved through a variety of methods: Optionally, the energy management system can use a hierarchical optimization method to first conduct macro-strategy planning and then refine it into specific control instructions. This method first determines the overall control direction based on the balance coefficient, then formulates detailed operating instructions for different types of equipment, and finally coordinates the execution order of each instruction. Optionally, the energy management system can also use a model predictive control method to generate a dynamic scheduling strategy through rolling optimization. This method dynamically adjusts control decisions by predicting the system state for multiple time steps in the future. It is understandable that other methods can also be used to generate management strategies, which are not limited here.

[0106] During the strategy generation process, energy management systems often encounter conflicting control objectives. To address this, the system employs a target prioritization mechanism: First, a multi-objective optimization framework is established, classifying objectives such as economic efficiency, reliability, and environmental protection according to their importance. A hierarchical decision-making approach is then employed to prioritize high-priority objectives, prioritizing those with lower priority. Finally, the Pareto optimality principle is employed to find the optimal balance among these multiple objectives. For example, the system prioritizes power supply reliability and then optimizes operating costs within this constraint.

[0107] In some embodiments, the energy management system will construct a fault warning and linkage control mechanism, that is, the energy management system will then obtain historical fault data of each energy gateway; the historical fault data includes fault type, fault time, fault duration and fault environment parameters; based on the energy management model, the historical fault data is correlated and analyzed to obtain equipment fault correlation feature data, and a fault warning threshold is generated based on the equipment fault correlation feature data; the equipment fault correlation feature data includes fault occurrence frequency, fault impact range and fault propagation path; the fault warning threshold includes operating parameter threshold, environmental parameter threshold and equipment status threshold; when it is detected that the operating data of any energy gateway exceeds the corresponding fault warning threshold, the linkage control data of the energy gateway group is generated, and the linkage control data is sent to the corresponding energy gateway; the linkage control data includes load transfer instructions, standby startup instructions and fault isolation instructions.

[0108] Historical fault data represents all fault-related information recorded by the energy gateway during operation, including a complete record of the fault type, time point, duration, and environmental parameters at the time of the fault. Equipment fault correlation feature data refers to fault characteristic patterns derived through data analysis, which characterize the correlation and propagation characteristics between faults. Fault warning thresholds represent the judgment criteria used to detect potential faults early, including limit settings for equipment operating parameters, environmental conditions, and status indicators. Linkage control data refers to the set of control instructions used for coordinated system response in the event of a fault.

[0109] After completing the training of the basic energy management model, the energy management system needs to establish a fault warning and prevention mechanism. Specifically, the energy management system first extracts fault records from the historical database of each energy gateway, including equipment fault alarms, abnormal operation records, maintenance records, etc. Then, using time series association analysis methods, it explores the temporal, causal, and spatial relationships between faults and establishes a fault propagation model. Based on this model, the system calculates the occurrence patterns and impact ranges of different types of faults, and sets multi-dimensional warning thresholds based on equipment reliability requirements. When it detects that real-time operating data exceeds the warning threshold, the system immediately generates linkage control instructions including load transfer, standby startup, and fault isolation, and sends them to the relevant equipment for execution through the energy gateway.

[0110] In some embodiments, fault warning analysis can be implemented in a variety of ways: Optionally, the energy management system can adopt a fault prediction method based on deep learning, capturing the evolution patterns of faults by constructing a time series prediction model. This method first extracts time series features from historical fault data, then establishes a fault prediction model through a recurrent neural network, and finally dynamically adjusts the warning threshold based on the prediction results. Optionally, the energy management system can also adopt a fault association analysis method based on a knowledge graph, mining fault patterns by establishing an equipment fault knowledge base. This method first constructs an equipment fault knowledge graph, then analyzes the fault propagation path based on graph calculations, and finally generates fault warning rules. It is understandable that other methods can also be used to implement fault warnings, which are not limited here.

[0111] During the fault warning process, the energy management system can encounter the problem of frequent fluctuations in warning signals, leading to false alarms. To address this, the energy management system employs a multi-dimensional fault analysis mechanism: first, it establishes multi-level warning trigger conditions, including the degree of parameter excursion, duration, and trend characteristics; then, it uses fuzzy reasoning to comprehensively assess the fault risk level; and finally, it implements appropriate control measures based on the risk level. For example, if a device's temperature is detected to be slightly above the limit, the system will first observe the trend of change and trigger coordinated control only if the temperature continues to rise and the impact expands.

[0112] S206: Monitor the online status of each energy gateway in the energy gateway group, and when it is detected that the target energy gateway is in an offline state, obtain historical operation data and associated gateway data of the target energy gateway.

[0113] The target energy gateway represents a specific gateway that is currently offline. Historical operation data refers to the gateway's operational records before it went offline. Associated gateway data includes the operational data of other gateways that interact with the target gateway, including the interacting energy type, interacting energy amount, and interaction time series data.

[0114] When the system detects a gateway offline, it activates an emergency response mechanism. Specifically, the energy management system first monitors the online status of each gateway in real time through a heartbeat detection mechanism. If a gateway is detected offline, the system immediately retrieves its historical operational data, including recent energy production and consumption data, equipment operating status, and more. Simultaneously, the system analyzes the energy interactions between the gateway and other gateways to determine the scope and extent of the impact.

[0115] In some embodiments, offline detection and data acquisition can be achieved in a variety of ways: Optionally, the energy management system can adopt a multi-level detection mechanism, combined with communication status detection, data update detection and business response detection, to improve the accuracy of offline judgment. This method avoids misjudgments and missed judgments by setting multiple detection dimensions and detection cycles; Optionally, the energy management system can also adopt a distributed status monitoring method to improve system reliability by monitoring each other by adjacent gateways. This method establishes a monitoring network between gateways to achieve rapid detection and location of faults. It is understandable that other methods can also be used to achieve offline monitoring, which is not limited here.

[0116] During offline detection, the energy management system can encounter false alarms due to fluctuations in communication quality. To address this, the system employs an offline judgment anti-jitter mechanism: First, a time threshold is set for offline judgment to prevent momentary communication interruptions from triggering an offline response. Then, the offline status is verified through multiple confirmations. Finally, the judgment parameters are dynamically adjusted based on historical communication quality data. For example, in areas with poor communication quality, the system will appropriately extend the judgment time to reduce the false alarm rate.

[0117] S207: Determine energy supply and demand characteristics of the target energy gateway based on historical operation data, and obtain target gateway characteristic data.

[0118] Energy supply and demand characteristics represent the typical patterns and behavioral characteristics of a gateway in energy production and consumption, including power generation characteristics (such as power generation timing distribution and power fluctuation characteristics), power consumption characteristics (such as load type and power consumption patterns), and energy storage characteristics (such as charging and discharging patterns and capacity utilization). Target gateway characteristic data is a digital description of these characteristics, including feature vectors, statistical parameters, and pattern indicators.

[0119] The energy management system requires in-depth analysis of the operating characteristics of offline gateways. Specifically, it first decomposes historical operating data into time series to extract daily, weekly, and seasonal cycle characteristics. Then, using statistical analysis methods, it calculates various characteristic parameters, such as average generated power, peak-to-valley differences, and load factors. The system also analyzes the gateway's abnormal event records, including fault characteristics and recovery time, to assess its reliability. Finally, the system integrates all characteristic data into a standardized feature vector, which serves as the basis for subsequent analysis.

[0120] In some embodiments, feature extraction can be achieved through a variety of methods: Optionally, the energy management system can use deep learning methods to automatically extract features, learning the implicit features of the data through an autoencoder network. This method first constructs a multi-layer encoder structure, extracts key features through dimensionality reduction learning, then verifies the validity of the features through reconstruction errors, and finally obtains a low-dimensional feature representation. Optionally, the energy management system can also use feature engineering methods based on expert knowledge to design feature extraction rules based on domain experience. This method comprehensively reflects the operating characteristics of the gateway by designing a multi-dimensional feature indicator system. It is understandable that other methods can also be used to achieve feature extraction, which are not limited here.

[0121] During feature extraction, energy management systems often encounter the problem of feature redundancy due to excessively high data dimensionality. To address this, the system employs a feature optimization mechanism: First, it reduces feature dimensionality through principal component analysis (PCA); then, it uses feature importance assessment methods, such as mutual information analysis, to identify key features; and finally, through feature combination, it constructs more explanatory composite features. For example, the system can combine multiple related time series features into a comprehensive metric to better represent the operational characteristics of a gateway.

[0122] S208: Calculate the impact coefficient of the target energy gateway being offline on the energy gateway group based on the target gateway characteristic data and the associated gateway data.

[0123] The impact coefficient is a comprehensive indicator used to quantify the impact of an offline event. It includes sub-items such as the energy supply impact coefficient, load impact coefficient, and network topology impact coefficient. Associated gateway data includes operational data for other gateways that have direct energy interactions with the offline gateway and is used to assess the scope of the impact.

[0124] The energy management system needs to accurately assess the scope and extent of the impact of an offline event. Specifically, it first constructs a diagram of energy flows between gateways, analyzing the offline gateway's position and role in the energy network. It then calculates the energy exchange volume of the gateway and assesses the direct supply-demand gap caused by its offline status. The system also analyzes the load-bearing capacity of associated gateways to assess the feasibility of energy reallocation. Finally, through comprehensive calculations, it derives multi-dimensional impact coefficients to guide subsequent policy adjustments.

[0125] In some embodiments, impact assessment can be achieved in a variety of ways: Optionally, the energy management system can use graph theory analysis methods to establish an energy network topology diagram to analyze the importance of nodes and the impact propagation path. This method first constructs a weighted directed graph model and then evaluates the influence of offline nodes through centrality analysis and flow analysis. Optionally, the energy management system can also use scenario simulation methods to derive impact assessment results through simulation calculations of multiple offline scenarios. This method simulates and analyzes system responses by setting up multiple hypothetical scenarios. It is understandable that other methods can also be used to achieve impact assessment, which are not limited here.

[0126] During the impact assessment process, the energy management system encounters the problem of the impact scope dynamically expanding. To address this, the energy management system employs a dynamic impact tracking mechanism: first, a multi-level impact propagation model is established to track the spread of impact; then, through real-time monitoring and updates, the impact assessment results are dynamically adjusted; finally, impact warning thresholds are set to promptly identify potential chain reactions. For example, if the load pressure of an associated gateway exceeds a threshold, the system will issue an early warning of possible secondary impacts.

[0127] S209 , updating parameters of the energy management model based on the influence coefficient to obtain an offline adjustment model, and regenerating the energy equipment management strategy based on the offline adjustment model.

[0128] The offline adjustment model is a new model created by adjusting the parameters of the original energy management model based on offline conditions. It includes revised prediction models, optimization models, and control models. The new management strategy includes emergency dispatch plans, load transfer strategies, and recovery plans.

[0129] The energy management system needs to quickly adjust its model to adapt to offline conditions. Specifically, it first determines the model parameters that need adjustment based on the influencing coefficients, such as load forecasting parameters and network topology parameters. It then updates these parameters using incremental learning methods to adapt the model to the new operating conditions. The system also regenerates the scheduling strategy based on the updated model, including power adjustment plans for the remaining gateways and load redistribution strategies. Finally, the system verifies the feasibility of the new strategy to ensure it meets system constraints.

[0130] In some embodiments, model updates can be achieved through a variety of methods. Optionally, the energy management system can employ online learning methods to continuously optimize model parameters using real-time data streams. This method establishes parameter update rules to dynamically adjust the model and maintain its timeliness. Alternatively, the energy management system can employ model migration methods, leveraging model experience from similar scenarios to rapidly construct new operational models. This method accelerates model adaptation through knowledge transfer. It is understood that other methods for model updates can also be employed, and these are not intended to be limiting here.

[0131] During the model update process, the energy management system may encounter unstable performance after the update. To address this, the energy management system adopts a progressive update mechanism: first, a sensitivity analysis of the model parameters is conducted to identify key parameters; then, a step-by-step update strategy is implemented to gradually adjust the model parameters; and finally, cross-validation is used to ensure the effectiveness of the update. For example, the system will first update the most influential parameters and then adjust other parameters after observing the system response.

[0132] In some embodiments, the energy management system will start a backup resource scheduling mechanism, that is, the energy management system will then obtain the load redundancy data of each energy gateway in the energy gateway group; the load redundancy data is used to characterize the backup energy capacity of each energy gateway; based on the load redundancy data, a target backup gateway is selected from the energy gateway group; the device startup timing and energy output curve of the target backup gateway are calculated to generate a backup startup instruction; the backup startup instruction includes the device startup time, energy output power and output duration; the backup startup instruction is sent to the target backup gateway, and the real-time operation data of the target backup gateway is obtained; the prediction parameters of the offline adjustment model are adjusted according to the real-time operation data, so that the deviation between the prediction result of the offline adjustment model and the real-time operation data is less than a preset error range.

[0133] The load redundancy data represents the excess energy capacity available for backup regulation at the energy gateway, including the backup capacity of power generation equipment and the available capacity of energy storage equipment. The target backup gateway refers to a specific gateway selected from the gateway group that possesses backup regulation capabilities. The backup activation command is a specific set of control commands used to activate the backup device.

[0134] The energy management system needs to quickly start backup resources when a gateway goes offline. Specifically, the energy management system first obtains the load status data of all gateways and calculates the backup capacity and response capability of each gateway. Then, based on evaluation indicators in multiple dimensions such as geographical location, response time, and adjustment cost, the optimal backup gateway is selected from gateways with sufficient redundancy. Based on the energy gap and load characteristics of the offline gateway, the system plans the startup process of the backup equipment in detail, including startup timing arrangement, power regulation curve and operation time setting. During the execution of the backup plan, the system collects operating data in real time, including power output, voltage frequency, equipment status, etc., and dynamically optimizes the parameter configuration of the prediction model based on the actual operating results.

[0135] In some embodiments, backup resource scheduling can be achieved in a variety of ways: Optionally, the energy management system can use a dynamic programming method to optimize the backup startup strategy. The method first establishes a multi-stage decision model that takes time series into account, and then gradually solves the optimal startup sequence based on system constraints, and finally generates a step-by-step startup plan; Optionally, the energy management system can also use a fast response mechanism to quickly call backup resources through a preset backup plan library. The method first establishes backup plans for a variety of typical scenarios, and then quickly selects a suitable plan through scenario matching, and finally fine-tunes the parameters. It is understandable that other methods can also be used to achieve backup scheduling, which is not limited here.

[0136] During the backup switchover process, the energy management system may encounter issues such as untimely response from the backup equipment, impacting system stability. To address this, the energy management system employs a progressive switchover mechanism: first, the dynamic response characteristics of the backup equipment are evaluated and startup parameters are appropriately set; then, a staged switchover strategy is implemented to avoid sudden load changes; and finally, real-time monitoring is used to ensure a smooth switchover process. For example, the system reserves sufficient warm-up time based on the equipment's startup characteristics and maintains appropriate power overlap during the switchover process.

[0137] S210 : Determine gateway characteristic data according to access information of the newly added energy gateway to the energy gateway group, and assign the newly added energy gateway to a corresponding gateway type group based on the gateway characteristic data.

[0138] Referring to step S103 , the energy management system groups the newly added energy gateways.

[0139] It's important to note that when assigning gateway types, the energy management system analyzes the gateway's communication parameters (including device address, baud rate, data bits, stop bits, parity, etc.) and communication protocol type (such as Modbus, Profibus, and CAN). The system uses a protocol conversion unit to uniformly process different protocols, ensuring that newly added gateways seamlessly integrate with existing systems.

[0140] S211. Generate simulation operation data of the newly added energy gateway according to the configuration information of the gateway type group.

[0141] Referring to step S104 , the energy management system generates simulation operation data.

[0142] It's important to note that when generating simulated operating data, the energy management system considers multiple operating scenarios, including varying weather conditions, load levels, and operating modes. This simulated data is then cleaned, denoised, and normalized using a preprocessing module to ensure data quality and usability.

[0143] In some embodiments, the energy management system will first execute a gateway verification process before simulating data, that is, the energy management system will first extract the device parameters and operating characteristics of the newly added energy gateway to obtain gateway verification data; the gateway verification data includes device model, specification parameters, power characteristics, communication protocol type and control mode; a verification rule set is constructed based on the gateway verification data, and the type of the newly added energy gateway is verified based on the verification rule set to obtain a verification result; when the verification result shows that there is a deviation, the type characteristic value of the newly added energy gateway is recalculated based on the gateway verification data.

[0144] Gateway verification data represents the basic characteristics of a newly added energy gateway, used to determine its positioning and functionality within the energy management system. The verification rule set is a standard set of rules used to determine a gateway's type. The type characteristic value is a quantitative indicator that characterizes the gateway's type attributes.

[0145] The energy management system needs to authenticate and determine the type of newly connected energy gateways. Specifically, the energy management system first collects detailed parameter information for the newly added gateway, including hardware configuration, performance parameters, and communication characteristics. Then, based on the existing gateway classification standards in the system, it establishes multi-dimensional verification rules that include device characteristics, functional characteristics, and performance characteristics. The system matches the feature data of the newly added gateway with the verification rules to determine its type. If any deviation is found in the type determination results, the system reanalyzes the gateway characteristics, adjusts the feature extraction method, and recalculates the type feature values.

[0146] In some embodiments, gateway verification can be achieved through various methods. Optionally, the energy management system can employ pattern recognition methods to determine gateway type. This method first extracts the gateway's multidimensional feature vector, then determines the characteristic patterns through cluster analysis, and finally determines the type based on similarity matching. Optionally, the energy management system can also employ an expert system approach to determine type. This method first establishes a rule-based reasoning system, then determines gateway attributes through a step-by-step process, and finally generates a type determination result. It is understood that other methods can also be employed to achieve gateway verification, and these are not limited here.

[0147] During the type verification process, the energy management system may encounter a mismatch between the characteristics of a new gateway and the existing classification standards. To address this, the energy management system employs an adaptive classification mechanism: first, feature analysis is performed to determine the gateway's key functional characteristics; then, the applicability of the existing classification system is evaluated; and finally, the classification standards are dynamically expanded as needed. For example, when a new energy storage device is connected, the system analyzes its unique charging and discharging characteristics and, if necessary, adds a new device type definition.

[0148] S212: Incrementally train the energy management model based on the simulated operation data to generate a simulated energy model.

[0149] Referring to step S105 , the energy management system generates a simulated energy model.

[0150] It should be noted that the energy management system employs a multi-stage training strategy, beginning with model pre-training and then fine-tuning using simulated data. During training, model performance is continuously optimized by adjusting parameters such as the learning rate and weight decay coefficient. Cross-validation is also used to evaluate the model's generalization capabilities.

[0151] S213 , adjusting the energy dispatching instructions and the operating parameter configurations based on the simulated energy model to obtain simulated dispatching instructions and simulated parameter configurations to adjust the energy equipment management strategy.

[0152] Referring to step S106 , the energy management system adjusts the energy device management strategy.

[0153] When adjusting its dispatch strategy, the system prioritizes the use of clean energy and optimizes it based on the regulatory capabilities of energy storage devices. For example, when sufficient solar power is predicted, the system prioritizes distributing solar energy to power-consuming devices and storing excess energy in storage batteries.

[0154] S214: Acquire actual operation data of the newly added energy gateway, determine simulation deviation parameters of the simulated operation data, and adjust the simulated energy model based on the simulation deviation parameters to obtain a target energy model.

[0155] Referring to step S107 , the energy management system determines a target energy model.

[0156] It's important to note that the energy management system incorporates the actual operating status and performance data of the equipment when adjusting its models. For example, for wind turbines, the system dynamically adjusts the prediction model parameters based on real-time wind speed and generator operating status. For energy storage batteries, the system considers the battery's health and charge-discharge characteristics when optimizing the model.

[0157] S215 , based on the target energy model, respectively modifying the simulated scheduling instructions and the simulated parameter configurations to actual scheduling instructions and actual parameter configurations to optimize the energy equipment management strategy.

[0158] Referring to step S108 , the energy management system optimizes the energy equipment management strategy.

[0159] It's important to note that the energy management system considers users' individual needs and price response preferences when optimizing management strategies. Through the user interface, the system receives user instructions for adjusting strategies, including optimization target configurations and energy usage preferences, and dynamically adjusts management strategies based on these inputs. The system also monitors strategy execution in real time, calculates user satisfaction indicators, and optimizes strategies as necessary.

[0160] In some embodiments, the energy management system can support user personalized configuration, that is, the energy management system can also receive policy adjustment instructions input by the user through the management terminal; the policy adjustment instructions include optimization target configuration, energy usage preferences, operating time period settings and cost control parameters; the optimization target configuration includes energy saving priority, cost priority, comfort priority and peak-valley balance priority; based on the policy adjustment instructions, a user personalized model is constructed, and the user personalized model is integrated with the target energy model for training to obtain a personalized energy model; based on the personalized energy model, the actual scheduling instructions and actual parameter configurations are optimized to obtain personalized scheduling instructions and personalized parameter configurations; the execution result data of the personalized scheduling instructions and personalized parameter configurations are monitored in real time, and the user satisfaction index is calculated based on the optimization target and execution result data set by the user; when the user satisfaction index is lower than the preset threshold during the continuous monitoring period, the personalized energy model is reconstructed.

[0161] The policy adjustment instruction represents the personalized energy management requirements entered by the user through the management terminal and is used to customize the energy management strategy. The optimization target configuration refers to the desired optimization direction set by the user, including priority settings for multiple dimensions such as energy saving, cost, comfort, and peak-valley balance. The personalized energy model represents a customized energy management model that incorporates the user's personalized needs. The user satisfaction index is an evaluation parameter that quantitatively assesses the degree to which energy management results align with user expectations.

[0162] Energy management systems need to dynamically adjust management strategies based on users' individual needs. Specifically, the energy management system first receives parameters such as optimization goals, energy usage preferences, operating time periods, and cost control settings set by users through the management terminal. Based on these personalized settings, it then constructs personalized models that reflect user needs, including energy usage habit models, comfort models, and cost sensitivity models. These personalized models are integrated and trained with the basic energy model to create customized energy models that adapt to users' individual needs. Based on this model, the system optimizes and adjusts existing scheduling instructions and parameter configurations to generate personalized management plans that meet user preferences. During plan execution, the system continuously monitors the results and calculates satisfaction metrics, including energy efficiency, cost savings, and comfort scores. If satisfaction metrics fall below preset standards for multiple consecutive monitoring periods, the system collects user feedback again and updates the personalized model.

[0163] In some embodiments, personalized strategy optimization can be achieved in a variety of ways: Optionally, the energy management system can adopt a multi-objective optimization method by constructing an optimization model that considers multiple optimization objectives. This method first converts the optimization objectives set by the user into mathematical constraints, then establishes a multi-objective optimization function for solution, and finally selects the best solution through Pareto optimality; Optionally, the energy management system can also adopt a reinforcement learning method to continuously optimize the decision-making strategy through interaction with the user. This method first establishes a reward mechanism for user feedback, then continuously improves the strategy through trial and error learning, and finally forms an optimal strategy that adapts to the user's preferences. It is understandable that other methods can also be used to achieve personalized optimization, which are not limited here.

[0164] During personalized optimization, energy management systems often encounter conflicting optimization objectives. To address this, the energy management system employs a dynamic balancing mechanism: First, a matrix of impact relationships among multiple objectives is constructed to assess their interactions. Then, based on user-defined priorities, the weighting coefficients of each objective are dynamically adjusted. Finally, through a rolling optimization approach, secondary objectives are gradually improved while maintaining the primary objective. For example, when a user simultaneously requests cost reduction and improved comfort, the system optimizes energy scheduling to reduce costs while maintaining basic comfort levels.

[0165] In the embodiments of the present application, due to the use of innovative technical solutions such as device feature recognition based on deep learning, fault correlation analysis based on knowledge graphs, and personalized optimization based on reinforcement learning, it is possible to achieve intelligent identification and classification of energy equipment, predictive prevention and control of faults, and personalized response to user needs, effectively solving technical problems such as difficulty in equipment adaptation, inaccurate fault warnings, and insufficient personalized management capabilities in traditional energy management systems, thereby achieving intelligent, personalized, and reliable management of energy systems. Specifically, the present application achieves rapid and accurate identification and classification of new equipment through multi-dimensional feature analysis, thereby improving the adaptability of the system; significantly improves the reliability of the system through fault correlation analysis and predictive prevention and control; and achieves customized energy management services through dynamic optimization of user personalized models; at the same time, the progressive optimization mechanism adopted by the present application enables the system to continuously accumulate experience and optimize management strategies, and has the ability to continuously improve.

[0166] The energy management system in the embodiment of the present invention is described below from the perspective of hardware processing. Figure 3 , is a schematic diagram of the structure of a physical device of the energy management system in an embodiment of the present application.

[0167] It should be noted that Figure 3The structure of the energy management system shown is only an example and should not limit the functions and scope of use of the embodiments of the present invention.

[0168] like Figure 3 As shown, the energy management system includes a CPU 301, which can execute various appropriate actions and processes based on programs stored in a ROM 302 or programs loaded from a storage unit 308 into a RAM 303, such as the methods described in the above embodiments. RAM 303 also stores various programs and data required for system operation. CPU 301, ROM 302, and RAM 303 are interconnected via a bus 304. An I / O interface 305 is also connected to bus 304.

[0169] The following components are connected to the I / O interface 305: an input section 306 including an audio input device, push button switches, and the like; an output section 307 including a liquid crystal display (LCD), an audio output device, indicator lights, and the like; a storage section 308 including a hard disk and the like; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card or a modem. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. Removable media 311, such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, is installed in the drive 310 as needed, so that computer programs read from the removable media can be installed in the storage section 308 as needed.

[0170] In particular, according to an embodiment of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, an embodiment of the present invention includes a computer program product comprising a computer program carried on a computer-readable medium, the computer program including a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 309 and / or installed from the removable medium 311. When the computer program is executed by the CPU 301, the various functions defined in the present invention are performed.

[0171] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. Each box in the flowchart or block diagram can represent a module, program segment, or part of the code, and the above-mentioned module, program segment, or part of the code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings.

[0172] Specifically, the energy management system of this embodiment includes a processor and a memory, and a computer program is stored in the memory. When the computer program is executed by the processor, the energy management method of collaborative AI and energy gateway provided in the above embodiment is implemented.

[0173] As another aspect, the present invention further provides a computer-readable storage medium, which may be included in the energy management system described in the above embodiments, or may exist independently and not be incorporated into the energy management system. The above storage medium carries one or more computer programs, which, when executed by a processor of the energy management system, enable the energy management system to implement the energy management method for collaborative AI and energy gateway provided in the above embodiments.

[0174] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0175] As used in the above embodiments, the term “when” may be interpreted to mean “if” or “after” or “in response to determining that” or “in response to detecting that”, depending on the context. Similarly, the phrases “upon determining that” or “if (stated condition or event) is detected” may be interpreted to mean “if determining that” or “in response to determining that” or “upon detecting (stated condition or event)” or “in response to detecting (stated condition or event)”, depending on the context.

Claims

1. An energy management method that collaborates with AI and energy gateway, characterized in that: Applied to an energy management system, the method includes: Obtain historical operating data of each energy gateway in the energy gateway group and generate an energy management model through machine learning algorithms; Based on the energy management model, generating an energy device management strategy for the energy gateway group including energy scheduling instructions and operating parameter configurations; Determining gateway characteristic data according to access information of the newly added energy gateway to the energy gateway group, and assigning the newly added energy gateway to a corresponding gateway type group based on the gateway characteristic data; generating simulation operation data of the newly added energy gateway according to the configuration information of the gateway type group; performing incremental training on the energy management model based on the simulated operation data to generate a simulated energy model; Adjusting the energy dispatching instruction and the operating parameter configuration respectively based on the simulated energy model to obtain simulated dispatching instruction and simulated parameter configuration to adjust the energy equipment management strategy; Acquiring actual operation data of the newly added energy gateway, determining a simulation deviation parameter of the simulated operation data, and adjusting the simulated energy model based on the simulation deviation parameter to obtain a target energy model; Based on the target energy model, the simulated scheduling instructions and the simulated parameter configuration are respectively modified to be actual scheduling instructions and actual parameter configuration, so as to optimize the energy equipment management strategy.

2. The method according to claim 1, characterized in that The step of generating an energy device management strategy for the energy gateway group including energy scheduling instructions and operating parameter configurations based on the energy management model specifically includes: Acquiring multidimensional operating data of each energy gateway in the energy gateway group; the multidimensional operating data includes power generation data, power consumption data, equipment efficiency data, meteorological environment data, and load distribution data; Inputting the multi-dimensional operation data into the energy management model to obtain time series prediction data of the energy gateway group; the time series prediction data includes a power generation time series curve, a power demand time series curve, and a load fluctuation time series curve; Calculating a load balancing coefficient of the energy gateway group based on the time series prediction data; the load balancing coefficient is used to represent the degree of matching between the power generation capacity and power demand of each energy gateway; According to the load balancing coefficient, energy dispatching instructions and operating parameter configuration data of the energy gateway group are generated to obtain an energy equipment management strategy.

3. The method according to claim 2, characterized in that After the step of generating energy dispatch instructions and operating parameter configuration data for the energy gateway group based on the load balancing coefficient to obtain an energy device management strategy, the method further includes: Obtaining historical fault data of each of the energy gateways; the historical fault data includes fault type, fault time, fault duration, and fault environment parameters; Performing correlation analysis on the historical fault data based on the energy management model to obtain equipment fault correlation feature data, and generating fault warning thresholds based on the equipment fault correlation feature data; the equipment fault correlation feature data includes fault occurrence frequency, fault impact range, and fault propagation path; the fault warning thresholds include operating parameter thresholds, environmental parameter thresholds, and equipment status thresholds; When it is detected that the operating data of any of the energy gateways exceeds the corresponding fault warning threshold, linkage control data of the energy gateway group is generated and sent to the corresponding energy gateway; the linkage control data includes load transfer instructions, standby startup instructions and fault isolation instructions.

4. The method according to claim 1, wherein After the step of respectively modifying the simulated scheduling instructions and the simulated parameter configuration to actual scheduling instructions and actual parameter configuration based on the target energy model to optimize the energy device management strategy, the method further includes: Receive a policy adjustment instruction input by a user through a management terminal; the policy adjustment instruction includes optimization target configuration, energy usage preference, operation time period setting, and cost control parameters; the optimization target configuration includes energy saving priority, cost priority, comfort priority, and peak-valley balance priority; Building a user personalized model based on the policy adjustment instruction, and fusing the user personalized model with the target energy model to obtain a personalized energy model; Optimizing the actual scheduling instruction and the actual parameter configuration based on the personalized energy model to obtain personalized scheduling instructions and personalized parameter configuration; monitoring the execution result data of the personalized scheduling instructions and the personalized parameter configuration in real time, and calculating a user satisfaction index based on the optimization goal set by the user and the execution result data; When the user satisfaction index is lower than a preset threshold during a continuous monitoring period, the personalized energy model is rebuilt.

5. The method according to claim 1, wherein After the step of generating an energy device management policy for the energy gateway group including energy scheduling instructions and operating parameter configurations based on the energy management model, the method further includes: Monitoring the online status of each energy gateway in the energy gateway group, and when detecting that a target energy gateway is offline, obtaining historical operation data and associated gateway data of the target energy gateway; the associated gateway data includes the gateway identification, interaction energy type, interaction energy amount, and interaction time series data of the target energy gateway; Determine the energy supply and demand characteristics of the target energy gateway based on the historical operation data to obtain target gateway characteristic data; Calculating, based on the target gateway characteristic data and the associated gateway data, an impact coefficient of the target energy gateway being offline on the energy gateway group; Parameters of the energy management model are updated based on the influence coefficient to obtain an offline adjustment model, and the energy equipment management strategy is regenerated based on the offline adjustment model.

6. The method according to claim 5, characterized in that After the steps of updating the parameters of the energy management model based on the influence coefficient to obtain an offline adjustment model, and regenerating the energy device management strategy based on the offline adjustment model, the method further includes: Obtaining load redundancy data of each energy gateway in the energy gateway group; the load redundancy data is used to characterize the backup energy capacity of each energy gateway; selecting a target backup gateway from the energy gateway group based on the load redundancy data; Calculate the device startup timing and energy output curve of the target backup gateway and generate a backup startup instruction; the backup startup instruction includes the device startup time, energy output power and output duration; Sending the standby startup instruction to the target standby gateway and obtaining real-time operation data of the target standby gateway; The prediction parameters of the offline adjustment model are adjusted according to the real-time operation data, so that the deviation between the prediction result of the offline adjustment model and the real-time operation data is less than a preset error range.

7. The method according to claim 1, characterized in that Before the step of generating the simulated operation data of the newly added energy gateway according to the configuration information of the gateway type group, the method further includes: Extracting the device parameters and operating characteristics of the newly added energy gateway to obtain gateway verification data; the gateway verification data includes device model, specification parameters, power characteristics, communication protocol type and control mode; Constructing a verification rule set according to the gateway verification data, and verifying the type of the newly added energy gateway based on the verification rule set to obtain a verification result; When the verification result shows that there is a deviation, the type characteristic value of the newly added energy gateway is recalculated based on the gateway verification data.

8. An energy management system, characterized in that: The energy management system includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the energy management system to execute the method described in any one of claims 1 to 7.

9. A computer-readable storage medium comprising instructions, characterized in that: When the instruction is executed on an energy management system, the energy management system is caused to execute the method according to any one of claims 1 to 7.

10. A computer program product, characterized in that When the computer program product is run on an energy management system, the energy management system is enabled to perform the method according to any one of claims 1 to 7.

Images