Smart energy digital twin management and control system and method

Abstract

The invention provides a smart energy digital twin management and control system and method, and the system comprises a control part, a data collection module, a communication and transmission module, an emergency response module and a server, and the server comprises a digital twin system which comprises optimal value substitution, active external cause progression and passive external cause learning. The method comprises the following steps: extracting real-time data from a sensor data set at any moment, determining an optimal value, setting a reference value at the next time, dividing amplitude information of a control value into a plurality of sub-amplitudes, transmitting each sub-amplitude to a control part in a subsequent time period according to an equal-score value or a decline value mode, and obtaining the optimal value of the control part based on a supervised learning method. Optimal value information extracted from sensor data in a previous time period is used as input data, and deep learning training is carried out by using a neural network. According to the scheme, various data of the energy system can be collected and analyzed in real time by using the digital twin technology, and a comprehensive and real-time energy efficiency analysis report is provided for managers, so that energy scheduling and control are optimized.

Description

Technical Field

[0001] The present invention belongs to the technical field of smart energy management, and specifically relates to a smart energy digital twin management and control system and method. Background Art

[0002] It has become an inevitable trend to combine the heating industry with information technology, replace the traditional model with emerging technologies, and promote the upgrading and transformation of the heating industry and realize the development of smart heating under the hard constraints of energy conservation and environmental protection. The disadvantages of the traditional model and the necessity of improving emerging technologies. At present, the energy digital twin management and control system uses advanced digital technology in the thermal industry to build a virtual copy of a physical thermal system, namely a "digital twin", through real-time data collection, sensor monitoring, data analysis and simulation technology. However, the traditional thermal industry management system still relies on manual scheduling and simple automation equipment. System monitoring and control are mainly based on basic sensor data, and often lack real-time dynamic optimization and intelligent decision-making. Many thermal companies use simple integrated control systems, which are limited to the control of single parameters such as temperature, flow, and pressure. They fail to make full use of big data and artificial intelligence technologies, and lack in-depth analysis and prediction of the overall performance and long-term operation trends of the system. The existing technology generally has the following shortcomings. Data islands: Data between various sensors and controllers are not fully shared, resulting in delayed or inaccurate information feedback, affecting the response speed and accuracy of the system. Lack of intelligent decision support: Traditional systems usually lack decision support based on big data and artificial intelligence, and cannot perform global optimization in complex environments, making it difficult to achieve optimal configuration of energy scheduling and thermal systems. Slow response: The system's slow response to the external environment (such as climate change, fluctuations in user demand, etc.) results in energy waste and reduced user comfort.

[0003] With the gradual increase in energy consumption and higher requirements for energy conservation and emission reduction, traditional thermal systems are facing problems such as low efficiency, energy waste and high management costs. Therefore, it is particularly important to adopt an intelligent management and control system based on digital twin technology. Summary of the invention

[0004] In view of the defects and problems existing in the prior art, the purpose of the present invention is to realize real-time monitoring, optimized control and adaptive feedback of smart energy systems through the combination of digital twin technology and deep learning algorithms, improve the accuracy, efficiency and intelligence of energy management, ensure the stable operation of the heating system, save energy and reduce consumption, and enhance user experience and system reliability.

[0005] The solution of the present invention to solve the technical problem is: adopt a smart energy digital twin control method, including a user-level heating control system and a system server, the user-level heating control system includes at least one closed-loop heating system, and the closed-loop heating system includes a sensor group, a control component and a digital twin system, the digital twin system includes a display part and a non-display part, the display part is responsible for real-time mapping of the sensor data set C to the corresponding pipelines and equipment in the digital twin system, and visually displaying the mapping results to users or managers through the display system; the non-display part is located inside the system for data processing; in the data processing process, the data set C={Q1,Q2,…,Qn} of the sensor component and the data set K={P1,P2,…,Pm} of the control component are taken, and the system server divides the system running time into k time periods, and the duration of each time period is Δt; the data processing process includes optimal value substitution, active external cause progression and passive external cause learning; optimal value substitution: at any time ti, from the sensor data set Ct{Q11…Q1t; Q21…Q2t;……Qn1…Qnt} extracts the real-time data Ci{Q1i; Q2i;……Qni} of each sensor from the real-time data Ci{Q1i; Q2i;……Qni} of each sensor, and determines the best value Cbest closest to the center value of the set threshold based on the deviation between the data of each sensor and the set threshold. The best value Cbest will be used as a reference for the setting value of each control component in the next time period ti+1. The setting information is transmitted to the control component through the transmission module, and the control component adjusts the working parameters based on the information; Active external factor progression: According to the amplitude of the set control value, the system divides the control information into multiple sub-amplitudes, and each sub-amplitude is transmitted to the control component of the subsequent time period in an equal or decreasing manner. Each sub-amplitude will be gradually and progressively applied as new setting information to the heating control of the subsequent time periods; Passive external factor learning: Based on the supervised learning method, the best value Cbest information extracted from the sensor data of the previous time period is used as the input data, and according to the heating fluctuation caused by external factors, the neural network is used for deep learning training, the model parameters are optimized by using the training set, and the model performance is evaluated by using the test set, so as to adaptively adjust and optimize the changes in external factors.

[0006] Preferably, it also includes unit-level heat supply control and regional-level heat supply control. The unit-level heat supply control connects multiple single closed-loop heating systems through a heating network to form a more centralized energy management network; each area of ​​the regional-level heat supply control includes multiple unit-level closed-loop systems, and multiple regional heating systems are connected in parallel for centralized scheduling and control.

[0007] A smart energy digital twin management and control system is also used, including at least one closed-loop heating system, control component: an actuator that adjusts the parameters of the heating system in real time according to sensor data, and the set value of the control component is calculated by the system server through data processing using the weight 1 method. The system server set value is transmitted to the control component through the transmission module to adjust its working state. Data acquisition and monitoring system: This system is used to collect sensor data and obtain data from each sensor in the system to form a data set; the data acquisition system receives data from the sensor and transmits it to the data layer, the system processes data through the business layer, generates control signals and passes them to the control component through the access layer, and the monitoring system also interacts with the display layer to display real-time status and warning information. Operation support layer: interacts with the data layer, business layer, and display layer through the infrastructure to ensure stable operation of the system. Communication and transmission module: responsible for ensuring the smooth flow of data from sensors to the data layer, from the data layer to the control components, and finally to the display layer. Data is transmitted from sensors to the data layer through the Internet of Things protocol. The data layer transmits the optimized control signal to the control component through the intranet or cloud service. The control signal is also transmitted to the control system through the network to achieve equipment adjustment. The display layer obtains real-time data and displays the system operation status and warning information. Operation and maintenance support hardware: including operation and maintenance terminals, operation and maintenance servers and emergency power supply hardware equipment, supporting system monitoring, early warning response, and abnormal situation processing. The operation and maintenance terminal receives warning signals from the display layer. System server and database: The server and database store historical data, real-time data and model results in the system. The sensor data is transmitted to the database for subsequent analysis and decision support. After the optimization algorithm is processed by the server, the calculated control signal is fed back to the control component. Emergency response and backup equipment: a backup mechanism used to start when a major failure occurs in the system. When a failure of the main device is detected, the signal will trigger the activation of the backup device. The system will automatically detect and reconfigure the control signal to enable the backup device to work normally.

[0008] The beneficial effects based on the above scheme include: 1. Real-time monitoring and precise control: Digital twin technology can collect and analyze various data of the energy system in real time, providing managers with comprehensive and real-time energy efficiency analysis reports, thereby optimizing energy scheduling and control.

[0009] 2. Intelligent scheduling and optimization: Through the introduction of deep learning algorithms, the system can adaptively adjust the energy distribution and temperature control strategies of each link, automatically handle complex energy demand fluctuations, and achieve precise scheduling and energy-saving optimization of the heating network.

[0010] 3. Energy saving and consumption reduction: The system can optimize heating and energy utilization efficiency and reduce energy waste according to user needs, environmental changes and historical data, thereby achieving significant energy saving effects and reducing operating costs.

[0011] 4. Improve system stability and reliability: Through digital twins and data-driven feedback mechanisms, it is possible to predict equipment failures, abnormal fluctuations, or load imbalances in advance, take preventive measures in advance, avoid sudden failures in the system, and improve system stability and reliability.

[0012] 5. Improve user experience: The system's intelligent temperature control and energy scheduling can provide a more comfortable user experience, while optimizing heat supply, avoiding energy waste, and providing users with more accurate heating services. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 It is a relationship block diagram of the single-level smart energy digital twin management and control system of the present invention; Figure 2 It is a relationship block diagram of a multi-level smart energy digital twin management and control system. DETAILED DESCRIPTION

[0014] The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0015] Example 1: A single-level smart energy digital twin management and control system, which is based on a user-level heating system. Its core goal is to use digital twin technology to monitor the energy system in real time, optimize control and provide data feedback to improve the accuracy, efficiency and intelligence of energy management. Through real-time monitoring, data analysis and intelligent decision-making, the digital twin management and control system can improve the operating efficiency of the system, reduce energy consumption, and provide more refined and personalized solutions for the thermal industry while ensuring the quality of heating. The following is an in-depth description of the solution, involving the functions and technical details of each part, such as Figure 1 shown.

[0016] 1. System Architecture Digital twin system: It is established through two-way feedback between the virtual model and the physical system, with the purpose of real-time mapping and controlling the state of the physical system. The system includes a display part and a non-display part. Display part: responsible for real-time mapping of sensor data with physical components such as system pipelines and equipment, and displaying the system operation status through visualization. For example, sensor data (such as temperature, pressure, flow, etc.) will be synchronized to the virtual model in real time, reflecting the status of the corresponding pipelines and equipment, so that staff can monitor and make decisions. Non-display part: mainly responsible for data processing and decision support, including data analysis and optimization tasks such as optimal value calculation, external factor progression, and deep learning. This part interacts directly with the control system to optimize system control through the simulation and prediction capabilities of the digital twin.

[0017] At the same time, the system architecture also includes different hierarchical definitions to ensure that each functional module can work together efficiently: Presentation layer (display part): The presentation layer is the interface between the system and the user. It is responsible for displaying the results of the business layer processing to the user and receiving the user's input and request. The presentation layer interacts with the business layer, obtains the results of the business layer processing, and displays them to the user. At the same time, it also receives the user's request through the user input and passes it to the business layer for processing. Access layer: The access layer is the interface for the system to exchange data with external users or systems. It is responsible for processing user requests and data transmission. The access layer interacts with the business layer, passes the user's request to the business layer for processing, and returns the processing results returned by the business layer to the user. At the same time, the access layer is also responsible for data exchange and integration with other systems. Business layer: The business layer is the core part of the system, responsible for processing specific business logic and realizing various business functions. It may include service monitoring, authentication module, device monitoring module, business processing module, etc. Each module or component is responsible for specific business functions. The business layer relies on the data support provided by the data layer and interacts with the user through the presentation layer. At the same time, it also communicates and collaborates with the operating environment through the operating support layer. Data layer: The data layer is built on the operation support layer and provides various information resources for the upper service layer (or business layer). It includes site data, equipment data, file data, space information, and system data storage. The data layer provides necessary data support for the business layer and is the basis for the business layer to implement various functions. At the same time, it also interacts with the operation environment through the operation support layer. Operation support layer: The operation support layer is located at the bottom of the architecture and is an important infrastructure supporting the entire system or project. It supports the current mainstream Kubernetes, docker, etc., and can achieve rapid deployment, portability and scalability of applications, simplifying the development, testing and deployment process of applications. The operation support layer provides the necessary operating environment and support for the data layer, business layer and presentation layer. It ensures the stability, reliability and security of the system. Operation environment: The system supports Linux or Windows cross-platform deployment and operation, including business systems, database systems, and middleware that also support operation in Linux and Windows environments. The operation environment provides necessary support and guarantee for the operation support layer, data layer, business layer and presentation layer. It is the basis for all layers to operate.

[0018] 2. Reaction system and control components Data collection and monitoring system: Real-time data collection (sensor group), monitoring and feedback mechanism of heating system are crucial. Through the Internet of Things, sensors and other equipment, data of various parameters are obtained, and intelligent adjustment and remote control are realized, which effectively improves the heating quality and operation efficiency of the system. User information management: timely obtain user heat consumption data, accurately measure, charge reasonably, and avoid excessive or insufficient heating.

[0019] In a single closed-loop heating system, the relevant parameters that affect the heating effect and efficiency mainly include the following aspects: temperature (temperature parameters): including water supply temperature, return water temperature and indoor temperature; pressure (pressure parameters): water supply pressure, return water pressure; flow (flow parameters): heating flow, return water flow; heat load: heating load and seasonal load changes; heat (heat transmission and utilization efficiency): heat loss and heat exchange efficiency; pipeline network parameters: pipeline layout and length and pipeline network sealing; environmental factors: outside temperature (air temperature) and wind speed and humidity; user needs: fluctuations in heat demand and heat dissipation of heat users; system operating status: system control and regulation capabilities and backup and emergency response capabilities; pipeline water quality.

[0020] Different types of sensors are used to monitor various parameters in the heating system, including temperature, pressure, flow, heat load, pipe network status, etc. in real time. By using various sensors, especially intelligent sensors combined with automatic control systems, efficient and accurate heating management can be achieved, the overall efficiency of the heating system can be improved, energy consumption can be reduced, user comfort can be ensured, and the goal of smart heating can be ultimately achieved.

[0021] The system operation time is divided into k time periods, each of which lasts for △t. At this time, the system collects, processes and analyzes sensor data in each time period, adjusts the behavior of the control components, and provides feedback optimization for the operation status of the entire system. In each time period of multiple consecutive time periods, the sensor data set Ct{Q11…Q1t; Q21…Q2t;…Qn1…Qnt}, C{Q1, Q2, ..., Qn}: This data has the characteristics of continuity and volatility; the sensor data set Ci{Q1i; Q2i;…Qni} at any time, where i∈t. Sensors are used to collect system operation data in real time to reflect the status of energy systems (such as heating systems). These sensors are usually deployed in different physical locations to measure key parameters such as temperature, pressure, and flow. n represents the number of sensors. Control components K{P1, P2, ..., Pm}: Various components of the control system, responsible for adjusting the system status in real time according to sensor data, such as regulating valves, fans, electric heaters, etc. The number of these control components is represented by m. By rationally selecting and adjusting various control components (such as temperature control valves, pressure reducing valves, flow control valves, etc.), the various parameters in the heating system are accurately adjusted, and dynamic adjustment is achieved through control equipment and control valves. According to the thresholds set for each sensor, the system obtains the optimal value Cbast of each sensor that is closest to the threshold (such as the center value of the threshold) in any ti period, and uses the optimal value Cbast as the setting value of each control component in the ti+1 period, and transmits the value to the server through the transmission module. The server provides the corresponding setting information to each control component for optimal value configuration.

[0022] 3. Data processing and control optimization Optimal value substitution: refers to calculating the optimal value Cbast close to the optimal threshold value according to the threshold and volatility of the sensor data in any time period ti, and using this value as the set value of the control component in the next time period. Due to the continuity and volatility of sensor data, the optimal value is not just a static value. If there is no active and passive external factors involved, the optimal value is used as a static setting value in the next time period △t. In theory, the setting value remains unchanged in the next time period △t. Since △t is short, the best overall performance is continuous dynamic. However, the setting value in the next time period △t usually changes dynamically, and the reason is that it is affected by active and passive external factors.

[0023] Calculation of the optimal value: At any time period ti, the system will compare the actual value of each sensor with the preset threshold value, for example, select the sensor data closest to the center value of the threshold as the "optimal value". The optimal value is a data set, which is a set of setting values ​​of a series of control components. Control component setting: The system transmits the calculated optimal value Cbast to the control component through the transmission module, and adjusts its working state to ensure that the energy system achieves the expected operating effect in the next time period.

[0024] Active external factor progression: refers to the optimization of artificial setting factors in the system (such as user settings, manual background adjustments, etc.). The changes in these external factors are usually random, and the adjustment range is arbitrary, so it is necessary to divide them into multiple sub-amplitudes according to the set amplitude, and adjust them step by step. Amplitude progression: Active external factor progression divides the adjustment amplitude into multiple stages according to the set amplitude, using equal or decreasing values. For example, if the user sets the temperature from 20°C to 25°C, the system divides this interval into multiple smaller increments and gradually applies these increments to the system. Through gradual adjustment, excessive system fluctuations are avoided and system stability is maintained. Sub-amplitude substitution: Each sub-amplitude is used as a new control parameter and is substituted into the subsequent time periods one by one to ensure that the control components can smoothly transition to the new set state.

[0025] Active exogenous learning (deep learning): In the process of active exogenous learning, the system optimizes according to the control effect of each period through the deep learning algorithm. The basic principle is to continuously adjust the behavior of the system through the reward mechanism so that the effect of each period is closer to the optimal state. Learning process: In each period, the system first uses the superposition value of the optimal value Cbast and the sub-amplitude as input, and predicts the impact of the set value on the system effect through the model. Then, the system compares the effect of the previous period with the setting of the current period to determine whether the current setting can improve the system performance. Reward mechanism: The system strengthens the optimal control strategy by setting a reward mechanism. When the control effect in the period is close to the expected target, the system gives positive rewards, otherwise it gives negative feedback. The deep learning algorithm adjusts the control strategy according to these reward signals to gradually optimize the operation effect of the entire system. Optimization algorithm: Based on algorithms such as reinforcement learning and deep learning, the system continuously improves the control strategy to achieve more intelligent self-regulation. By learning historical data and real-time feedback, the system can make more accurate predictions and adjustments in future periods.

[0026] Passive external factors usually refer to abnormal events caused by external environment or equipment failure in the digital twin system. Such factors usually have no direct control or active intervention, but are system problems caused by environmental changes or the operating status of the equipment itself. Common passive external factors include pipe network failure, valve failure, sensor failure and line failure, which are characterized by suddenness and randomness.

[0027] In the digital twin system, supervised learning methods are usually used to learn about passive external factors, and the identification and prediction of abnormal events are achieved through the construction of training sets and test sets. Specifically, the training set contains a large amount of historical data related to passive external factors, and the test set is used to evaluate the generalization ability and prediction performance of the training model. The whole process is summarized in the following steps.

[0028] Before model training, the best value Cbast data in the previous period needs to be preprocessed, including denoising, missing value filling and feature engineering. By extracting features related to passive external factors (such as equipment status, environmental changes, sensor readings, etc.) from historical data, high-quality data support is provided for subsequent learning. Constructing loss function: For the prediction problem of passive external factors, the warning signal is usually optimized as a loss function. The loss function is defined as the error measure between the model prediction value and the actual target value. For example, if the prediction signal fails to reflect the abnormal situation in time, the value of the loss function is large, otherwise it is small. Optimizing the loss function helps the model gradually adjust the parameters to improve the prediction accuracy. After the model training is completed, the test set needs to be used for performance verification. The test set should contain different types of passive external factor data. The model compares the prediction of these data with the actual results to evaluate the generalization ability and accuracy of the model. By analyzing the performance of the model on the test set, potential problems are found and optimized. The optimization process may include further adjusting the network structure, selecting a suitable activation function, adjusting the learning rate, etc. In addition, combined with the feedback mechanism, real-time operation data is fed back to the training set to continuously improve the model's predictive ability.

[0029] In practical applications, this process can help the digital twin system monitor and identify abnormal situations caused by passive external factors in real time, and then issue early warning signals in time to ensure the safety and stability of system operation. By combining deep learning methods, it can effectively deal with failures caused by various external factors, discover potential risks in time, and provide decision support for pipeline network, valve group and equipment maintenance.

[0030] Internal warning and external warning: The warning system caused by passive external factors is subdivided into internal warning and external warning. The two have different functions and roles in the digital twin system, and their specific application scenarios and display mechanisms are different.

[0031] Internal early warning mainly refers to the activation of the self-repair mechanism when the system encounters an abnormal situation caused by passive external factors. This type of early warning is limited to internal use of the system and does not display any warning signals to the outside world. After receiving the abnormal signal caused by passive external factors, the system will recover to normal operation by adjusting control parameters, enabling backup equipment, or performing automated repair operations. If the system can successfully recover or reduce the impact of the abnormality in the internal mechanism and does not affect the stability or function of the external system, there is no need to issue early warning information to the outside world. Internal early warning is limited to system self-diagnosis and repair, and is not displayed through external interfaces or user interfaces. Automatic repair through internal feedback loops and algorithms prevents small-scale abnormalities from gradually evolving into large-scale faults. If the system can resume normal operation through self-repair, no human intervention is required. Application scenarios: Applicable to systems that can automatically identify and repair minor faults or fluctuations, such as automatic calibration of abnormal fluctuations in sensor data, automatic cleaning of local blockages in pipelines, etc.

[0032] External warning refers to when the digital twin system encounters a more serious passive external factor, the system still cannot recover to the optimal or normal state after self-repair, and then needs to warn the operation and maintenance personnel or decision makers through an external warning mechanism. At this time, the warning information will be presented to the external system or operator for further manual intervention or optimization measures. The key feature of external warning is that the system has not been able to recover to the ideal state after facing an abnormality, and this state of failure to fully recover will pose a potential threat to the safety, reliability or efficiency of the system. Therefore, the system will mark the severity of the current state through different levels of warning signals (such as yellow and red warnings).

[0033] Yellow warning: Yellow warning usually means that the system has deviated from the optimal state and there are potential risks, but it has not yet caused a serious impact on the overall system function. At this time, the operating status of the system has exceeded the set normal range, but through further adjustment or monitoring, the occurrence of failures can still be avoided. Application scenarios: such as high pipeline pressure, slow valve response speed, sensor data fluctuations, etc. Although the system can partially recover after self-repair, it has not yet returned to the optimal working state. Impact: The yellow warning shows the outside world that the system may need to be monitored or is about to enter the optimization stage, reminding relevant personnel to pay attention.

[0034] Red alert: A red alert means that the system has not been able to recover to normal working state through self-repair and is close to or has reached the critical point of failure. At this time, the operation of the system may affect the entire production process or have caused serious threats to equipment and production safety. Application scenarios: For example, serious pipeline failures, complete valve failures, complete failures of key sensors, etc., even through the self-repair process, they have not recovered to normal state and there is a high risk. Impact: Red alerts require immediate response, which usually means that external operation and maintenance personnel are required to intervene to troubleshoot and repair the problem to avoid accidents or increased losses.

[0035] Labeling in the digital twin system: Once the system determines that the current state falls within the external warning range, the digital twin system will issue a warning through color labeling (such as yellow or red). The system presents this labeling to relevant operation and maintenance personnel or managers through a visual interface to help them quickly understand the current state and its possible impact. Warning information feedback mechanism: In addition to displaying color labels in the system interface, the digital twin system should also support sending warning information to relevant personnel through email, SMS or other communication channels, with necessary context data and recommended operations for quick response.

[0036] 4. Expandable part Through the visual interface, users can view the operating status and energy consumption data of each energy system in real time. The system can automatically detect abnormal energy consumption and inefficient equipment and provide optimization suggestions.

[0037] Environmental data visualization module: Use wireless sensors to collect indoor and outdoor temperature, humidity, CO2 concentration, VOC concentration, PM2.5 concentration and other environmental variables. According to weather classification, the system can achieve self-update in different environments and propose energy storage optimization solutions such as photovoltaic and wind power. When facing disastrous weather such as heavy rain and strong winds, the management department can issue instructions in the first time and take energy supply countermeasures.

[0038] Energy consumption information visualization module: Display key indicators such as air conditioning, lighting and electricity consumption statistics of special channels through the 2D visualization panel on the left. Connect to the background data in real time, monitor the building energy consumption, and display the contribution statistics of energy conservation and emission reduction. Provide various analysis reports based on the database to provide a reliable basis for energy-saving optimization and transformation.

[0039] Power supply equipment visualization module: includes sensors, smart IoT GPRS / 4G / NB-LOT, and big data system platform equipment information, enabling 2D configuration effects that can be viewed on mobile phones anytime and anywhere.

[0040] Power distribution system module: Based on a building power supply system, it simulates the application scenario of power distribution equipment in a building. The entire process of power supply from the underground power distribution room to each floor and then to each device is expressed through the flow of lines. Each device in the power distribution room can respond to interactive events, and when tripping, short circuit, leakage, etc. occur, an alarm signal can be issued in time.

[0041] Water supply system module: Import CAD drawings into the system and restore the secondary water supply pump room in proportion. Achieve unified management of water supply flow, pump frequency, inlet and outlet pressures, and water turbidity, pH value, residual chlorine, etc. Solve problems such as changes in water demand and pump stoppage due to accidents in the actual operation of the water supply system to ensure normal operation.

[0042] Example 2: Multi-level smart energy digital twin management and control system: The core goal is to use digital twin technology to monitor the energy system in real time, optimize control and data feedback to improve the accuracy, efficiency and intelligence level of energy management. In large-scale smart energy systems, multi-closed-loop series-parallel heating systems (for example, large-scale heating systems in industrial parks, communities and other units) are optimized through a hierarchical and modular management architecture. Multi-regional parallel heating systems (for example, heating systems in multiple industrial parks, communities or administrative areas, etc.) are comprehensive management of multiple multi-closed-loop series-parallel heating systems. In order to achieve global optimization, dynamic adjustment and intelligent feedback of the system, intelligent management and control based on information collection, control and feedback mechanisms are essential. Deep learning algorithms will further enhance the intelligent management level of the system, and through efficient data processing, prediction and optimization control, the entire heating network will be able to maximize energy efficiency, reduce operating costs, and improve heating quality.

[0043] The following is the management architecture of a multi-zone parallel heating system based on a deep learning algorithm, covering a detailed description of information collection, control, feedback, and optimization mechanisms.

[0044] 1. System architecture and hierarchical design (such as Figure 2 shown) The design of the smart energy digital twin management and control system is divided into multiple levels. The goal of each level is to carry out refined management according to different heating areas, and achieve efficient operation of the overall system through layer-by-layer optimization.

[0045] 1.1 User level (single closed-loop heating system) The user level is the most basic single closed-loop heating system, such as the heating system of a single home, office or factory. Each user system monitors environmental conditions in real time through sensors and adjusts local control devices (such as thermostats, regulating valves, pumps, etc.) according to demand to ensure optimal indoor temperature and energy utilization.

[0046] 1.2 Unit level (multiple closed-loop series and parallel heating systems) At the unit level, multiple single closed-loop heating systems (such as multiple residences, office buildings, industrial plants, etc.) are connected through a heating network to form a more centralized energy management network. The system goal at this level is to optimize the heat distribution and load scheduling between single systems.

[0047] 1.3 Regional level (multi-region parallel heating system) The regional level involves the overall management of multiple heating systems in multiple industrial parks, communities or administrative areas. Each region may consist of multiple unit-level closed-loop systems, and multiple regional heating systems are connected in parallel for collaborative management. Its main goal is to improve the heating efficiency and energy utilization of the entire region through global coordination and optimization.

[0048] 2. Multi-level smart energy digital twin management and control 2.1 Multi-layer features User-level overall energy demand data set: The total heating energy demand, total energy loss, total power consumption, total water consumption, and total heat supply cost of each user level in the unit level are obtained from the display part of the digital twin system. Overall scheduling control data set: total heating supply, total water supply, total output power, output and input temperature, output and input pressure, total pipe flow, and total energy efficiency of each unit level. Overall external factor data set: User behavior factors at each unit level (heating demand of office buildings during the day and at night, heating demand of residents during the day and at night), external environmental factors (climate change and seasonal differences, outdoor temperature changes), time factors (commuting factors, holidays and special circumstances), system configuration and maintenance factors (equipment performance and maintenance status, accuracy of intelligent control systems), system structure and design factors (building thermal performance, system configuration and pipeline layout). Population mobility issues must also be considered at the regional level.

[0049] Regarding the unit level (multiple closed-loop series-parallel heating system) and the regional level (multiple regional parallel heating system), there are many control and coordination management processes between user levels (single closed-loop heating system). Therefore, when managing and scheduling different user levels, the "overall energy demand data set", "overall scheduling control data set" and "overall external factor data set" of each user level are managed in a unified manner.

[0050] 2.2 Data processing at the unit level and regional level The unit level or regional level includes the overall energy demand data set and the overall scheduling control data set, and the system operation time is divided into k time periods, each of which lasts for Δt; different management and scheduling modules at the unit level or regional level include user-level optimal value substitution, management level control scheduling and system learning; user-level optimal value substitution: extract the energy consumption demand data group of each user level at any time from the continuous data of the overall energy consumption demand data set, and determine the optimal energy consumption demand data group for each user level through multi-objective optimization (for example, using a non-dominated sorting genetic algorithm to find a set of solutions with good balance among multiple demand data in the overall energy consumption demand data set, which is suitable for scenarios where the demand data are contradictory or mutually constrained), management level control scheduling: use the optimal energy consumption demand data group as the scheduling control data of the unit level or regional level in the next time period, and transmit this setting information to the master control unit at the unit level through the transmission module. The master control unit at the unit level adjusts the operation of each control component at the user level based on the information; system learning: based on the supervised learning method, the optimal energy consumption demand data group is used as the input data, the overall external factor data set is used as the loss function, and the neural network is used for deep learning training. The model parameters are optimized by using the training set, and the model performance is evaluated by using the test set, so as to adaptively adjust and optimize the changes in external factors.

[0051] The above specific embodiments of the present invention are only used to illustrate or explain the principles of the present invention, and do not constitute a limitation of the present invention. Therefore, any modification, equivalent substitution, improvement, etc. made without departing from the spirit and scope of the present invention should be included in the protection scope of the present invention.

Claims

1. A smart energy digital twin management and control method, including a user-level heating management and control system and a system server, characterized in that: The user-level heating control system includes at least one closed-loop heating system, and the closed-loop heating system includes a sensor group, a control component and a digital twin system. The digital twin system includes a display part and a non-display part. The display part is responsible for mapping the sensor data set C to the corresponding pipelines and equipment in the digital twin system in real time, and visually displaying the mapping results to users or managers through the display system; the non-display part is located inside the system for data processing; during the data processing process, the data set C={Q1,Q2,…,Qn} of the sensor component and the data set K={P1,P2,…,Pm} of the control component are taken, and the system server divides the system operation time into k time periods, and the duration of each time period is Δt; the data processing process includes optimal value substitution, active external factor progression and passive external factor learning; optimal value substitution: at any time ti, extract the real-time data Ci{Q1i; Q21; Q1t; Qn1; Qnt} of each sensor from the sensor data set Ct{Q11; Q21; Q2t; Qn1; Qnt} Q2i;……Qni}, and based on the deviation of each sensor data from the set threshold, determine the optimal value Cbest closest to the central value of the set threshold, the optimal value Cbest will be used as a reference for the setting value of each control component in the next time period ti+1, and the setting information will be transmitted to the control component through the transmission module, and the control component will adjust the working parameters based on the information; Active external factor progression: According to the amplitude of the set control value, the system divides the control information into multiple sub-amplitudes, and each sub-amplitude is transmitted to the control component of the subsequent time period in an equal or decreasing manner, and each sub-amplitude will be used as new setting information and gradually applied to the heating control of the subsequent time periods; Passive external factor learning: Based on the supervised learning method, the optimal value Cbest information extracted from the sensor data of the previous time period is used as the input data, and according to the heating fluctuations caused by external factors, the neural network is used for deep learning training, the model parameters are optimized by using the training set, and the model performance is evaluated by using the test set, so as to adaptively adjust and optimize the changes in external factors.

2. The smart energy digital twin management and control method according to claim 1 is characterized in that: It also includes unit-level heat supply control and regional-level heat supply control. The unit-level heat supply control connects multiple single closed-loop heating systems through a heating network to form a more centralized energy management network. Each area of ​​the regional-level heat supply control is composed of multiple unit-level closed-loop systems, and multiple regional heating systems are connected in parallel for centralized scheduling and control.

3. The smart energy digital twin management and control method according to claim 1 is characterized in that: It also includes active exogenous learning: During the active exogenous learning process, the system optimizes the control effect of each time period through a deep learning algorithm, and continuously adjusts the system's behavior through a reward mechanism, so that the effect of each time period is closer to the optimal state; Learning process: In each time period, the system first uses the superposition value of the optimal value Cbast and the sub-amplitude as input, and predicts the impact of the set value on the system effect through the model. Then, the system will compare the effect of the previous time period with the setting of the current time period to determine whether the current setting can improve the system performance; Reward mechanism: The system strengthens the optimal control strategy by setting a reward mechanism. When the control effect in the time period is close to the expected target, the system gives positive rewards, otherwise it gives negative feedback. The deep learning algorithm will adjust the control strategy according to these reward signals to gradually optimize the operation effect of the entire system; Optimization algorithm: Based on algorithms such as reinforcement learning and deep learning, the system can continuously improve the control strategy to achieve more intelligent autonomous adjustment. By learning historical data and real-time feedback, the system can make more accurate predictions and adjustments in future time periods.

4. The smart energy digital twin management and control method according to claim 1 is characterized in that: It also includes internal warning signals and external warning mechanisms. The internal warning signal is when encountering an abnormal situation caused by passive external factors, the self-repair mechanism is activated first. This type of warning is limited to internal use of the system and does not display any warning signals to the outside world. After receiving the abnormal signal caused by passive external factors, the system will adjust the control parameters, enable backup equipment, or perform automated repair operations to restore to normal operating state. If the system can successfully recover or reduce the impact of the abnormality in the internal mechanism and does not affect the stability or function of the external system, there is no need to issue warning information to the outside world; the external warning mechanism is when the digital twin system encounters a more serious passive external factor impact, the system still cannot be restored to the optimal or normal state after self-repair, and then it is necessary to warn the operation and maintenance personnel or decision makers through the external warning mechanism.

5. The smart energy digital twin management and control method according to claim 4 is characterized in that: Once the system determines that the current status falls within the external warning range, the digital twin system will issue a warning through color marking. The system will present this marking to relevant operation and maintenance personnel or managers through a visual interface to help them quickly understand the current status and its possible impact. The digital twin system supports sending warning information to relevant personnel via email, SMS or other communication channels, and attaches necessary contextual data and recommended operations for quick response.

6. The smart energy digital twin management and control method according to claim 2 is characterized in that: The unit-level heating control is used to optimize the heat distribution and load scheduling between single systems. Each single closed-loop system collects temperature, pressure, and flow parameters through sensors and uploads the data to the central control platform. Through the central control platform, the system adjusts the load distribution of each unit according to the collected data to ensure uniform heat distribution and avoid excess or insufficient heat in certain areas. The control platform is dynamically adjusted according to actual load and user needs to ensure stable operation and efficient energy consumption of each closed-loop system.

7. The smart energy digital twin management and control method according to claim 2 is characterized in that: The regional heating control is globally coordinated and optimized. The heating system in each region collects data in real time through the sensor network, including the temperature, pressure, flow, load demand, etc. of the region. These data are transmitted to the central system platform for global data analysis. The overall control system dispatches the heating load of each region based on the global data, and the system adjusts the heating strategy of each region in real time according to actual feedback.

8. A smart energy digital twin management and control system, comprising at least one closed-loop heating system, characterized in that: Control component: an actuator that adjusts the parameters of the heating system in real time according to the sensor data. The set value of the control component is calculated by the system server through data processing using the weighted method. The set value of the system server is transmitted to the control component through the transmission module to adjust its working state. Data acquisition and monitoring system: This system is used to collect sensor data and obtain data from each sensor in the system to form a data set; the data acquisition system receives data from the sensor and transmits it to the data layer. The system processes the data through the business layer, generates control signals and transmits them to the control components through the access layer. The monitoring system also interacts with the display layer to display real-time status and warning information; Operation support layer: interacts with the data layer, business layer, and presentation layer through the infrastructure to ensure stable operation of the system; Communication and transmission module: responsible for ensuring the smooth flow of data from sensors to the data layer, from the data layer to the control components, and finally to the display layer. Data is transmitted from sensors to the data layer through the Internet of Things protocol. The data layer transmits the optimized control signal to the control component through the intranet or cloud service. The control signal is also transmitted to the control system through the network to realize equipment adjustment. The display layer obtains real-time data and displays the system operation status and early warning information.

9. A smart energy digital twin management and control system according to claim 8, characterized in that: Including emergency response and backup equipment: a backup mechanism that is activated when a major system failure occurs. When a main device failure is detected, the signal will trigger the activation of the backup device. The system will automatically detect and reconfigure the control signal to enable the backup device to work normally.

Images