Building energy management system and method suitable for multiple energy sources

Abstract

The invention provides a building energy management system and method suitable for multiple energy sources, and relates to the technical field of energy management, and the method comprises the steps: constructing a multi-energy coupling map through the multi-dimensional information flow of the building energy sources; respectively carrying out distributed acquisition and source load disturbance modeling on state data of each energy node in the multi-energy coupling map to obtain an energy disturbance set; determining an adjustment tolerance interval through a scheduling strategy network based on energy efficiency evaluation and an energy disturbance set, and performing strategy action reconstruction on the multi-dimensional information flow through the adjustment tolerance interval to obtain an attenuation constraint term; penalty correction is carried out on the fluctuation state of the multi-dimensional energy flow in the building according to the attenuation constraint term, and an offset calibration factor is obtained; according to the energy supply and demand data of the building and the offset calibration factor, the output stream of the SCADA system is periodically evolved and updated to obtain the optimal scheduling strategy stream of the building energy, and the multi-energy collaborative optimization and strategy evolution of the multi-energy building system can be realized so as to improve the energy utilization efficiency.

Description

Technical Field

[0001] This application relates to the field of energy management technology, and more specifically, to a building energy management system and method applicable to multiple energy sources. Background Technology

[0002] With the increasing complexity of urban building energy consumption structures and the advancement of the "dual carbon" strategy, building energy management technology has received widespread attention as an important support for achieving energy conservation, carbon reduction, and intelligent operation of buildings. Energy management involves monitoring, analyzing, and optimizing the energy flow within buildings to improve energy efficiency while ensuring comfort, and it is an important component of modern building operation.

[0003] Traditional building energy management methods primarily target the independent control of single energy systems, making them ill-suited to the complex structures of modern buildings where multiple energy sources—electricity, heat, cooling, gas, water, and renewable energy—coexist and operate in a coupled manner. To address the multi-energy operation requirements of buildings, energy management technologies such as rule-based multi-energy scheduling strategies, model predictive control-based energy flow optimization methods, and multi-objective heuristic algorithms have emerged, achieving a degree of coordinated optimization of multi-energy flows. However, existing energy management technologies still rely on fixed control logic, making it difficult to adjust strategies based on energy source and load status, and failing to consider the impact of source and load uncertainty and fluctuations on strategy stability. In current energy management technologies, SCADA systems serve only as underlying data acquisition tools, lacking deeply integrated intelligent control logic for scheduling strategies, thus hindering the realization of multi-energy coordinated optimization in buildings. Therefore, how to achieve multi-energy coordinated optimization and strategy evolution in multi-energy building systems to improve energy efficiency remains a significant challenge for the industry. Summary of the Invention

[0004] This application provides a building energy management system and method applicable to multiple energy sources, which can realize multi-energy collaborative optimization and strategy evolution of multi-energy building systems to improve energy utilization efficiency.

[0005] In a first aspect, this application provides a building energy management system and method applicable to multiple energy sources, the management method comprising the following steps: Based on the SCADA system, multi-dimensional information flow of building energy is obtained, and then a multi-energy coupling map is constructed from the multi-dimensional information flow. The state data of each energy node in the multi-energy coupling graph are collected in a distributed manner and source-load disturbance modeling is performed to obtain the energy disturbance set; The adjustment tolerance range is determined by the scheduling strategy network based on energy efficiency assessment and the energy disturbance set. Then, the multidimensional information flow is reconstructed by the adjustment tolerance range to obtain the attenuation constraint term. The fluctuation state of multidimensional energy flow inside the building is collected, and the fluctuation state is penalized and corrected according to the attenuation constraint term to obtain the offset calibration factor. Based on the building's energy supply and demand data and the offset calibration factor, the output stream of the SCADA system is periodically updated to obtain the building's optimized energy scheduling strategy stream.

[0006] In this embodiment, acquiring the multi-dimensional information flow of building energy based on the SCADA system specifically includes: Collect different types of energy data using sensor terminals; All energy data are preprocessed based on the SCADA system to obtain a preprocessed energy dataset; The preprocessed energy dataset is extracted through the data interface to obtain a multidimensional information flow.

[0007] In this embodiment, constructing a multi-energy coupling map from the multi-dimensional information flow specifically includes: The multidimensional information stream is classified and aggregated, and then the temporal coupling characteristics between different types of energy are determined from the energy data subsets obtained by classification and aggregation. Determine the graph structure of the multi-energy coupling spectrum; The graph structure is weighted by all temporal coupling features to obtain a multi-energy coupling graph.

[0008] In this embodiment, the state data of each energy node in the multi-energy coupling graph are collected in a distributed manner and source-load disturbance modeling is performed to obtain the energy disturbance set, which specifically includes: The status data of each energy node in the multi-energy coupling graph is obtained based on a distributed database. The electrical and thermal load of each energy node is predicted based on its status data to obtain the electrical and thermal load fluctuation of each energy node. The electrical and thermal load fluctuations and status data of each energy node are aggregated to obtain an energy disturbance set.

[0009] In this embodiment, determining the adjustment tolerance range through the energy efficiency assessment-based scheduling strategy network and the energy disturbance set specifically includes: The energy disturbance set is input into the scheduling strategy network based on energy efficiency assessment for strategy evaluation to obtain the allocation strategy of electric and thermal power; Determine energy efficiency indicators, and then construct a multi-objective evaluation model based on the energy efficiency indicators; The tolerance evaluation of the electrothermal power allocation strategy is carried out using the multi-objective evaluation model to obtain the adjustment tolerance range.

[0010] In this embodiment, the attenuation constraint term obtained by reconstructing the multidimensional information flow through the adjustment tolerance interval specifically includes: Different types of energy tolerance thresholds are determined by the aforementioned adjustment tolerance range; Extracting state datasets from multidimensional information streams; The state dataset is filtered by state policy based on all energy tolerance thresholds to obtain attenuation constraint terms.

[0011] In this embodiment, the fluctuation state of multidimensional energy flow inside the building is collected using a sliding time window algorithm.

[0012] In this embodiment, the SCADA system is an energy data acquisition and monitoring control system based on a strategy evolution mechanism.

[0013] In this embodiment, the scheduling strategy network based on energy efficiency assessment is a cooperative control network constructed based on deep deterministic strategy gradient.

[0014] Secondly, this application provides a building energy management system suitable for multiple energy sources, for implementing a building energy management method suitable for multiple energy sources, the management system comprising: The information coupling module is used to acquire multi-dimensional information flow of building energy based on the SCADA system, and then construct a multi-energy coupling map from the multi-dimensional information flow. The disturbance sensing module is used to perform distributed acquisition of the state data of each energy node in the multi-energy coupling spectrum and source-load disturbance modeling, thereby obtaining the energy disturbance set; The strategy reconstruction module is used to determine the adjustment tolerance range through the scheduling strategy network based on energy efficiency assessment and the energy disturbance set, and then reconstruct the strategy action of the multidimensional information flow through the adjustment tolerance range to obtain the attenuation constraint term. The fluctuation correction module is used to collect the fluctuation state of multidimensional energy flow inside the building, and to penalize and correct the fluctuation state according to the attenuation constraint term to obtain the offset calibration factor. The evolution update module is used to periodically evolve and update the output stream of the SCADA system based on the building's energy supply and demand data and the offset calibration factor, so as to obtain the building's energy optimization scheduling strategy stream.

[0015] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects: The system acquires multi-dimensional information flow of building energy based on the SCADA system, and then constructs a multi-energy coupling map from the multi-dimensional information flow. Distributed acquisition and source-load disturbance modeling are performed on the state data of each energy node in the multi-energy coupling map to obtain an energy disturbance set. An adjustment tolerance range is determined through a scheduling strategy network based on energy efficiency assessment and the energy disturbance set. The multi-dimensional information flow is then reconstructed using the adjustment tolerance range to obtain an attenuation constraint term. The fluctuation state of the multi-dimensional energy flow within the building is collected, and the fluctuation state is penalized and corrected according to the attenuation constraint term to obtain an offset calibration factor. The output flow of the SCADA system is periodically updated based on the building's energy supply and demand data and the offset calibration factor to obtain an optimized scheduling strategy flow for building energy.

[0016] Therefore, this application demonstrates the ability to achieve multi-energy collaborative optimization and strategy evolution for multi-energy building systems. Firstly, by acquiring multi-dimensional information flow of building energy based on a SCADA system and constructing a multi-energy coupling map, multi-dimensional information fusion and energy flow graph structure modeling enhance the identification capability of building energy coupling characteristics and the overall control foundation, providing a unified data semantic platform for subsequent multi-energy coordinated scheduling. Furthermore, by distributively collecting energy node states and constructing source-load disturbance models, an energy disturbance set is generated, enabling quantitative modeling and scenario simulation of multi-source-load uncertainties. This improves the management system's perception and prediction capabilities of energy disturbances, providing a comprehensive disturbance input basis for scheduling strategy optimization. Secondly, by employing a scheduling strategy network based on energy efficiency assessment combined with the disturbance set to set adjustment tolerance intervals and reconstructing scheduling actions, optimization is achieved. This approach effectively avoids over-adjustment or inaction issues in strategy output. By constructing attenuation constraints, it achieves quantitative penalty feedback on the impact of scheduling actions on the stability and energy efficiency of the management system, improving the controllability of energy scheduling. Then, by collecting energy fluctuation status and generating offset calibration factors based on attenuation constraints, it can self-adjust based on real-time operational deviations of energy flow, which helps enhance the robust response capability of the management system to load fluctuations and external disturbances. Finally, by periodically evolving and updating the SCADA output stream based on supply and demand data and offset calibration factors, an optimized scheduling strategy stream is formed, constructing a closed-loop learning mechanism between the strategy network and the management system's operating status. This helps improve the self-optimization capability of the SCADA system, thereby realizing the adaptive evolution of intelligent scheduling strategies for building energy management.

[0017] In summary, the technical solution adopted in this application can realize multi-energy collaborative optimization and strategy evolution of multi-energy building systems to improve energy utilization efficiency. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this embodiment of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart of a building energy management method applicable to multiple energy sources provided in this application; Figure 2 This is an exemplary flowchart for determining the set of energy disturbances according to the present application; Figure 3 This is an exemplary flowchart for determining the adjustment tolerance range according to the present application; Figure 4 This is a module structure diagram of the management system provided in this application. Detailed Implementation

[0020] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] This application provides a building energy management system and method applicable to multiple energy sources. Its core is based on acquiring multi-dimensional information flow of building energy through a SCADA system, and then constructing a multi-energy coupling graph from the multi-dimensional information flow. Distributed acquisition and source-load disturbance modeling are performed on the state data of each energy node in the multi-energy coupling graph to obtain an energy disturbance set. An adjustment tolerance range is determined through a scheduling strategy network based on energy efficiency assessment and the energy disturbance set, and then the multi-dimensional information flow is reconstructed using the adjustment tolerance range to obtain an attenuation constraint term. The fluctuation state of the multi-dimensional energy flow within the building is collected, and the fluctuation state is penalized and corrected according to the attenuation constraint term to obtain an offset calibration factor. The output flow of the SCADA system is periodically updated based on the building's energy supply and demand data and the offset calibration factor to obtain an optimized scheduling strategy flow for building energy.

[0022] Example 1: To better understand the above technical solution, the following will provide a detailed description of the technical solution in conjunction with the accompanying drawings and specific implementation methods. (Refer to...) Figure 1 As shown, this figure is an exemplary flowchart of a building energy management method applicable to multiple energy sources, according to this embodiment of the present application. The management method includes the following steps: In step S1, a multi-dimensional information flow of building energy is obtained based on the SCADA system, and then a multi-energy coupling map is constructed from the multi-dimensional information flow.

[0023] It should be noted that the SCADA system in this application is an energy data acquisition and monitoring control system based on a strategy evolution mechanism. The strategy evolution mechanism refers to the dynamic learning and adaptive adjustment of historical data, real-time feedback and environmental parameters of the SCADA system's operating status to generate update rules for key control strategies such as data acquisition frequency, alarm triggering conditions and communication bandwidth allocation, thereby improving the SCADA system's perception sensitivity and scheduling response capability to complex energy systems.

[0024] In this embodiment, the acquisition of multi-dimensional information flow of building energy based on the SCADA system can be specifically carried out in the following manner: Collect different types of energy data using sensor terminals; All energy data are preprocessed based on the SCADA system to obtain a preprocessed energy dataset; The preprocessed energy dataset is extracted through the data interface to obtain a multidimensional information flow.

[0025] In practice, firstly, sensor terminals deployed inside the building collect different types of energy data in real time. This energy data includes: electrical data (such as voltage, current, and power), air conditioning system data (such as cooling load and supply / return air temperature and humidity), lighting system data (such as circuit power consumption and brightness levels), and water resource data (such as flow rate, pressure, and temperature). The sensor terminals may include electricity meters, water meters, temperature and humidity sensors, flow sensors, and smart switches. The installation locations of these sensor terminals can be determined based on the building's electrical distribution diagram and actual functional areas. Then, the PLC or edge computing nodes within the SCADA system perform data cleaning, time alignment, and normalization on all the energy data to obtain a pre-processed energy dataset. Finally, the pre-processed energy dataset is extracted using the standardized Modbus TCP data interface protocol to obtain a multi-dimensional energy information data set containing time-series electrical parameters, thermal parameters, water resource parameters, and other parameters. This energy information dataset is then transformed into a multi-dimensional information flow of building energy, which can be used for data modeling, condition diagnosis, and scheduling optimization tasks.

[0026] It should be noted that the multidimensional information flow in this application refers to a structured dataset that integrates the operating parameters of various energy systems within a building, with time series as the main axis. This structured dataset can comprehensively reflect the spatial distribution characteristics and dynamic trends of energy use in a building. In addition, the preprocessing operation of energy data using the SCADA system in this embodiment not only improves the availability and stability of the original data, but also unifies the semantics and accuracy of data from different sources by synchronizing timestamps and normalizing formats.

[0027] In this embodiment, the construction of the multi-energy coupling map from the multi-dimensional information flow can be specifically carried out in the following manner: The multidimensional information stream is classified and aggregated, and then the temporal coupling characteristics between different types of energy are determined from the energy data subsets obtained by classification and aggregation. Determine the graph structure of the multi-energy coupling spectrum; The graph structure is weighted by all temporal coupling features to obtain a multi-energy coupling graph.

[0028] In specific implementation, firstly, a hierarchical clustering algorithm is used to classify and aggregate energy data, including grouping parameter data according to energy type (such as electricity, cooling, heating, and water), resulting in energy data subsets such as electricity information subset, air conditioning system information subset, and water resource information subset. Then, canonical correlation analysis is used to extract the correlation coefficients of the time series of the electricity information subset, air conditioning system information subset, and water resource information subset. Finally, the ratio of the correlation coefficients between different types of energy data subsets is used as the temporal coupling characteristic between these different types of energy data subsets. Through the above method, the temporal coupling characteristics between different types of energy can be obtained, and these temporal coupling characteristics can represent the energy coupling behavior between different energy systems. Then, different types of energy data subsets are used as nodes in the graph, and the correlation coefficients of the time series of different types of energy data subsets are used as edges in the graph. Each edge connects two nodes, resulting in a graph structure of a multi-energy coupling spectrum. Preferably, an undirected weighted graph modeling method can be used, where each node represents a type of energy, and the weight of the edge represents the coupling strength between two nodes. Finally, weights are assigned to the edges in the graph structure. All time-series coupling features can be normalized, and then the normalized time-series coupling features are used as the weight values ​​of the edges in the graph structure. Thus, the graph structure with assigned weight values ​​is used as an energy coupling spectrum, which can quantitatively express the interaction strength between various energy systems.

[0029] It should be noted that the multi-energy coupling graph in this application refers to a graph model that represents the coupling relationship and interaction characteristics between various energy systems in a building in a time series in the form of a graph structure. The multi-energy coupling graph can reveal the dependence relationship of energy systems such as electricity, cooling, heating, and water in actual operation, and can reflect the energy transfer path and dynamic linkage mode between systems. In addition, the typical correlation analysis algorithm used in this embodiment can visualize the complex coupling relationship by capturing the potential coupling rules of different energy systems in the operation sequence. By introducing the edge weight adjustment mechanism of the graph structure, it is beneficial to realize the transition from data-driven to structural modeling.

[0030] In step S2, the state data of each energy node in the multi-energy coupling graph are collected in a distributed manner and source-load disturbance modeling is performed to obtain the energy disturbance set.

[0031] Preferably, in this embodiment, reference Figure 2 As shown, this diagram is an exemplary flowchart for determining the energy disturbance set according to the present application. In this embodiment, the state data of each energy node in the multi-energy coupling spectrum is collected in a distributed manner and source-load disturbance modeling is performed to obtain the energy disturbance set. This can be achieved by the following steps: In step S21, the status data of each energy node in the multi-energy coupling graph is obtained based on the distributed database; In step S22, the state data of each energy node is used to predict the electrical and thermal load, thereby obtaining the electrical and thermal load fluctuation of each energy node; In step S23, the electrical and thermal load fluctuations and status data of each energy node are aggregated to obtain an energy disturbance set.

[0032] In specific implementation, firstly, a key-value index is established for each energy node in the multi-energy coupling graph using a distributed database. The state data of each energy node is then extracted using this key-value index, yielding the state data for each energy node in the multi-energy coupling graph. This state data includes energy consumption, load power, temperature control parameters, voltage and current, and operating time. Next, a support vector regression model is used to predict the electrical and thermal load of each energy node's state data. Specifically, the state data of each energy node is used as input to the support vector regression model, and the output of the model yields the electrical and thermal load fluctuations of each energy node. Finally, the electrical and thermal load fluctuations and state data of each energy node are vectorized using one-hot encoding to obtain a feature vector for each energy node. This feature vector includes the average fluctuation amplitude, rate of change, and probability of abrupt change in the load prediction, used to describe the energy disturbances of the energy node. The set of feature vectors from all energy nodes is considered the energy disturbance set.

[0033] It should be noted that the energy disturbance set in this application refers to a set of multi-dimensional disturbance features that integrate the predicted results of electric and heat loads with the current state information, used to reflect the operational volatility of multi-energy coupled networks in buildings; in addition, the distributed database used in this embodiment is an energy consumption data management platform that supports time-series features and has high throughput, high availability and horizontal scalability; the source-load disturbance modeling in this application refers to the process of quantitatively modeling the uncertainty caused by load fluctuations, equipment switching or energy efficiency changes between various energy supply and consumption ends in buildings.

[0034] In step S3, the adjustment tolerance range is determined by the scheduling strategy network based on energy efficiency assessment and the energy disturbance set. Then, the multidimensional information flow is reconstructed by the adjustment tolerance range to obtain the attenuation constraint term.

[0035] It should be noted that the scheduling strategy network based on energy efficiency assessment in this application is a collaborative control network constructed based on deep deterministic policy gradient, including: a policy network (Actor), a value assessment network (Critic), an action selection module, and an energy efficiency reward feedback mechanism. The collaborative control strategy network, constructed using the deep deterministic policy gradient algorithm in deep reinforcement learning, has the core function of adaptively outputting control actions for multi-source devices under dynamic disturbance environments based on the state information of the building energy system, thereby achieving global energy efficiency optimization and system stability control. Preferably, the scheduling strategy network in this application adopts a dual-network structure (i.e., actor-critic architecture), where the actor network is used to generate scheduling actions, and the critic network is used to evaluate the value of the actions, which helps improve policy optimization efficiency and training stability. Furthermore, the training process of the scheduling strategy network is constructed using the policy optimization framework TensorFlow, and the training samples come from dynamic combinations of energy disturbance sets, enabling end-to-end intelligent control policy learning.

[0036] Preferably, in this embodiment, reference Figure 3 As shown, this figure is an exemplary flowchart for determining the adjustment tolerance range according to the present application. In this embodiment, determining the adjustment tolerance range through the scheduling strategy network based on energy efficiency assessment and the energy disturbance set can be achieved by the following steps: In step S31, the energy disturbance set is input into the scheduling strategy network based on energy efficiency assessment for strategy evaluation to obtain the allocation strategy of electric and thermal power; In step S32, energy efficiency indicators are determined, and then a multi-objective evaluation model is constructed based on the energy efficiency indicators; In step S33, the tolerance evaluation of the electrothermal power allocation strategy is performed through the multi-objective evaluation model to obtain the adjustment tolerance range.

[0037] In specific implementation, a test environment for strategy evaluation is constructed. This involves inputting an energy disturbance set into the energy efficiency-based scheduling strategy network to simulate the building energy system's operating state under different energy disturbance conditions. The strategy network then outputs an allocation strategy for electrothermal power, where the allocation strategy includes control parameters for multiple energy units such as fuel cells, heat pumps, and energy storage devices. Next, energy efficiency indicators are set to measure the system's operating performance. Preferably, these indicators include parameters such as primary energy utilization rate, overall system energy consumption, carbon dioxide emission intensity, and final-state energy storage offset. Based on these energy efficiency indicators, a multi-objective evaluation model is constructed using weighted integration, giving the model a unified evaluation function. Finally, the multi-objective evaluation model is used to evaluate the allocation strategy's performance, and an interval analysis is performed on the allocation values. Specifically, based on the target performance variation range, a threshold is defined that allows for small fluctuations in the strategy output. The allocation value plus the threshold is used as the upper limit of the interval, and the allocation value minus the threshold is used as the lower limit, resulting in an adjustment tolerance interval. The allocation values ​​include energy allocation values ​​for fuel cells, heat pumps, and energy storage devices.

[0038] It should be noted that the adjustment tolerance range in this embodiment refers to the acceptable fluctuation range of the scheduling strategy output under given disturbance conditions and energy efficiency constraints, without triggering over-limit behavior. It is used to reflect the flexibility of the building energy system regulation and ensure that the system achieves a dynamic balance between performance, economy and safety. The multi-objective evaluation model takes multi-dimensional energy efficiency indicators as input to ensure that the generated adjustment tolerance range has both the safety guarantee of system operation and the directional constraint driven by energy efficiency optimization. Preferably, in this embodiment, the construction and normalization evaluation process of the multi-objective evaluation model can be realized by using the scikit-learn library in Python, and the monitoring and updating of the adjustment tolerance range can be realized by using NumPy and Pandas.

[0039] In this embodiment, the attenuation constraint term is obtained by reconstructing the strategy action of the multidimensional information flow through the adjustment tolerance interval, which can be specifically done in the following way: Different types of energy tolerance thresholds are determined by the aforementioned adjustment tolerance range; Extracting state datasets from multidimensional information streams; The state dataset is filtered by state policy based on all energy tolerance thresholds to obtain attenuation constraint terms.

[0040] In specific implementation, firstly, different types of energy tolerance thresholds are extracted from the adjustment tolerance range output by the multi-objective evaluation model. These energy tolerance thresholds include multiple dimensions such as the output power tolerance threshold of fuel cells, the operating frequency tolerance threshold of heat pumps, and the charging and discharging power tolerance threshold of energy storage systems. Preferably, the energy tolerance thresholds can be represented by dynamic intervals, for example, using the current strategy output value ±Δ as the upper and lower limits, where Δ can be set according to the operating characteristics of different devices. Then, a state dataset is extracted from the multi-dimensional information flow obtained from the SCADA system. Finally, the extracted state dataset is compared item by item with each of the above energy tolerance thresholds, and a state strategy filtering mechanism is used to screen the effectiveness of actions. That is, for the strategy output corresponding to the state variable that exceeds the tolerance threshold, the strategy is reconstructed. Preferably, the strategy reconstruction can adopt a lightweight adjustment method based on gradient descent to compress the strategy output value into the tolerance range, so that the reconstructed action can replace the original strategy output to participate in subsequent scheduling execution. The joint dataset consisting of all state variables that have undergone strategy reconstruction and their reconstruction strategies can be used as a decay constraint term.

[0041] It should be noted that the attenuation constraint term in this application refers to the set of restrictive constraint parameters introduced for state variables exceeding the tolerance threshold during the strategy output process. The role of the attenuation constraint term is to dynamically suppress the regulation behavior that deviates too much from the energy efficiency, ensuring that the strategy output meets the actual energy equipment operation constraints while ensuring the stable operation of the system. In addition, the strategy action reconstruction in this application refers to the constrained reparameterization of action variables in the strategy network output that do not meet the adjustment tolerance range limit, thereby ensuring that the scheduling behavior meets the requirements of system stability and safe operation of equipment. The action reconstruction and state strategy filtering process can be implemented based on the SciPy library in Python to implement matrix operations and reconstruction judgment logic, which has good portability and integration efficiency. In this embodiment, the reconstruction strategy can be implemented by local first-order gradient search, which is beneficial to balance computational efficiency and strategy feasibility.

[0042] In step S4, the fluctuation state of multidimensional energy flow inside the building is collected, and the fluctuation state is penalized and corrected according to the attenuation constraint term to obtain the offset calibration factor.

[0043] In this embodiment, the fluctuation state of multidimensional energy flow inside the building is collected using a sliding time window algorithm.

[0044] It should be noted that the fluctuation state of multidimensional energy flow in this application refers to the instantaneous fluctuation and changing trend of different types of energy flow in a building energy system within a certain time interval. The energy flow includes, but is not limited to, multiple dimensions such as electrical energy flow, heat energy flow, gas flow and energy storage flow. The fluctuation state of multidimensional energy flow inside the building can be collected by a sliding time window algorithm. The sliding time window algorithm is a dynamic sampling algorithm that performs continuous local statistics on the state data of various energy flows inside the building within a fixed time span. It can dynamically capture the local fluctuation characteristics in energy data, thereby forming a subset of input information that can be used for energy consumption analysis, state determination and strategy optimization.

[0045] In this embodiment, the offset calibration factor is obtained by penalizing and correcting the fluctuation state according to the attenuation constraint term, which can be done in the following way: Map the fluctuation state to the fluctuation vector space; By jointly modeling the attenuation constraint term and the fluctuation vector space, the degree of energy storage offset caused by the fluctuation state of different energy types can be obtained. The offset calibration factor is obtained by weighting all energy storage offsets according to the preset penalty weights.

[0046] In specific implementation, firstly, the multidimensional energy flow fluctuation state collected within the sliding time window is standardized into a numerical expression, and then projected onto a low-dimensional fluctuation vector space using a principal component analysis algorithm. This fluctuation vector space is used to uniformly characterize the disturbance direction and intensity of different energy types. Secondly, a joint modeling function is constructed, that is, the aforementioned attenuation constraint term is mapped and weighted to the disturbance vector in the fluctuation vector space term by term to obtain the degree of energy storage offset caused by each energy disturbance to the energy storage system. Specifically, a two-layer neural network model can be used to complete the coupling process. The degree of energy storage offset refers to the deviation of the charge state from the target state caused by the multidimensional energy flow disturbance inside the building. Finally, based on the response delay, adjustment inertia, and power amplitude of different energy devices on the energy storage system, a set of penalty weight coefficients is preset, and the degree of energy storage offset caused by various energy disturbances is weighted and aggregated. The weighted aggregation result is used as the offset calibration factor. The weights in the weighted aggregation process can be set according to the control sensitivity of different energy types and the response elasticity coefficient of the energy storage system to various disturbances.

[0047] It should be noted that the offset calibration factor in this application is a correction parameter that reflects the degree of deviation of the charge state of the energy storage system caused by the disturbance state of different types of energy. Its essence is a compensation coefficient based on the fluctuation response, which can dynamically correct the state prediction error caused by scheduling actions and improve the robustness of energy balance modeling. In this embodiment, by fusing the attenuation constraint term with the fluctuation vector space, it is possible not only to capture the energy storage offset trend caused by short-term disturbances, but also to set differentiated penalty weights according to the physical characteristics of different types of energy. The penalty weights refer to the weighted parameters used to measure the severity of the impact of disturbances on the energy storage system.

[0048] In step S5, the output stream of the SCADA system is periodically updated based on the building's energy supply and demand data and the offset calibration factor to obtain the building's optimized energy scheduling strategy stream.

[0049] In this embodiment, the output stream of the SCADA system is periodically updated based on the building's energy supply and demand data and the offset calibration factor to obtain the optimized energy scheduling strategy stream for the building. Specifically, this can be achieved in the following manner: Collect energy supply and demand data for the building during the current scheduling cycle; The energy supply and demand data are input into the SCADA system, and iterative retraining is performed based on the offset calibration factor to obtain updated strategy parameters; The updated strategy parameters are redeployed to the output interface of the SCADA system, and then the optimized scheduling strategy flow is output through the output interface.

[0050] In practical implementation, energy supply and demand data refers to a set of data used to describe the quantitative relationship between energy supply and energy demand in a building's energy system at a specific moment or within a specific period. This includes energy supply data and energy demand data. Energy supply data includes: the power generation and heat output of fuel cells, the operating load and heat output of heat pumps, the instantaneous power generation output and available capacity of renewable energy sources (such as photovoltaics and wind power), and energy purchases. Energy demand data includes: the building's load-side electricity demand (lighting, equipment, air conditioning, etc.), building heat load demand (heating, hot water supply, refrigeration recovery, etc.), peak-valley load characteristics and their predicted values, and user behavior data. This data is used for instantaneous and periodic load estimation under the influence of factors such as air conditioning start-stop modes and elevator usage frequency. It should be noted that energy supply and demand data are usually expressed in time series form and are dynamically updated according to the scheduling cycle set by the SCADA system. They have multi-dimensional, multi-time scale, and strong uncertainty characteristics, and are the key basic data source for constructing the input features of the scheduling strategy network, evaluating strategy deviations, and generating optimized scheduling schemes. The energy supply and demand data can be collected and updated through edge terminals, smart meters, heat meters, sensor networks, and load forecasting models, and then transmitted to the SCADA system through standardized interfaces for centralized processing and strategy decision-making.

[0051] It should be noted that the optimized scheduling strategy flow in this application refers to a sequence of building energy scheduling instructions dynamically generated by integrating historical feedback data and strategy evolution mechanisms. It has time-scale adaptability and strategy iteration capability, and can effectively respond to changes in energy disturbances and system state shifts. The SCADA system can adopt the strategy gradient optimization algorithm in deep reinforcement learning, and perform goal-oriented training on the strategy network through the feedback of supply and demand data to enhance its energy efficiency regulation capability under multiple scenarios and multiple source load disturbances. In addition, the deployment of the updated scheduling strategy network adopts a modular hot replacement method to ensure uninterrupted operation of the SCADA system and improve the timeliness of strategy updates and the stability of system operation.

[0052] Example 2: This application provides a building energy management system suitable for multiple energy sources, referencing... Figure 4 As shown in the figure, this is a module structure diagram of the management system according to this embodiment of the present application. The management system includes: The information coupling module 100 is used to acquire multi-dimensional information flow of building energy based on the SCADA system, and then construct a multi-energy coupling map from the multi-dimensional information flow. The disturbance sensing module 200 is used to perform distributed acquisition of the state data of each energy node in the multi-energy coupling spectrum and source-load disturbance modeling, thereby obtaining an energy disturbance set; The strategy reconstruction module 300 is used to determine the adjustment tolerance range through the scheduling strategy network based on energy efficiency assessment and the energy disturbance set, and then reconstruct the strategy action of the multidimensional information flow through the adjustment tolerance range to obtain the attenuation constraint term. The fluctuation correction module 400 is used to collect the fluctuation state of the multidimensional energy flow inside the building, and to penalize and correct the fluctuation state according to the attenuation constraint term to obtain the offset calibration factor. The evolution update module 500 is used to periodically evolve and update the output stream of the SCADA system based on the building's energy supply and demand data and the offset calibration factor, so as to obtain the building's energy optimization scheduling strategy stream.

[0053] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0054] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compactdisc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0055] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

Claims

1. A building energy management method applicable to multiple energy sources, characterized in that, The management method includes the following steps: Based on the SCADA system, multi-dimensional information flow of building energy is obtained, and then a multi-energy coupling map is constructed from the multi-dimensional information flow. The state data of each energy node in the multi-energy coupling graph are collected in a distributed manner and source-load disturbance modeling is performed to obtain the energy disturbance set; The adjustment tolerance range is determined by the scheduling strategy network based on energy efficiency assessment and the energy disturbance set. Then, the multidimensional information flow is reconstructed by the adjustment tolerance range to obtain the attenuation constraint term. The fluctuation state of multidimensional energy flow inside the building is collected, and the fluctuation state is penalized and corrected according to the attenuation constraint term to obtain the offset calibration factor. Based on the building's energy supply and demand data and the offset calibration factor, the output stream of the SCADA system is periodically updated to obtain the building's optimized energy scheduling strategy stream.

2. The building energy management method applicable to multiple energy sources as described in claim 1, characterized in that, The multi-dimensional information flow of building energy acquired based on SCADA systems specifically includes: Collect different types of energy data using sensor terminals; All energy data are preprocessed based on the SCADA system to obtain a preprocessed energy dataset; The preprocessed energy dataset is extracted through the data interface to obtain a multidimensional information flow.

3. The building energy management method applicable to multiple energy sources as described in claim 1, characterized in that, The construction of a multi-energy coupling map from the multi-dimensional information flow specifically includes: The multidimensional information stream is classified and aggregated, and then the temporal coupling characteristics between different types of energy are determined from the energy data subsets obtained by classification and aggregation. Determine the graph structure of the multi-energy coupling spectrum; The graph structure is weighted by all temporal coupling features to obtain a multi-energy coupling graph.

4. The building energy management method applicable to multiple energy sources as described in claim 1, characterized in that, The state data of each energy node in the multi-energy coupling graph are collected in a distributed manner and source-load disturbances are modeled to obtain the energy disturbance set, which specifically includes: The status data of each energy node in the multi-energy coupling graph is obtained based on a distributed database. The electrical and thermal load of each energy node is predicted based on its status data to obtain the electrical and thermal load fluctuation of each energy node. The electrical and thermal load fluctuations and status data of each energy node are aggregated to obtain an energy disturbance set.

5. A building energy management method applicable to multiple energy sources as described in claim 1, characterized in that, Determining the adjustment tolerance range through the energy efficiency assessment-based scheduling strategy network and the energy disturbance set specifically includes: The energy disturbance set is input into the scheduling strategy network based on energy efficiency assessment for strategy evaluation to obtain the allocation strategy of electric and thermal power; Determine energy efficiency indicators, and then construct a multi-objective evaluation model based on the energy efficiency indicators; The tolerance evaluation of the electrothermal power allocation strategy is carried out using the multi-objective evaluation model to obtain the adjustment tolerance range.

6. The building energy management method applicable to multiple energy sources as described in claim 1, characterized in that, By reconstructing the strategy action of the multidimensional information flow through the aforementioned adjustment tolerance interval, the attenuation constraint term is obtained, specifically including: Different types of energy tolerance thresholds are determined by the aforementioned adjustment tolerance range; Extracting state datasets from multidimensional information streams; The state dataset is filtered by state policy based on all energy tolerance thresholds to obtain attenuation constraint terms.

7. A building energy management method applicable to multiple energy sources as described in claim 1, characterized in that, The fluctuation state of multidimensional energy flow inside the building is collected using a sliding time window algorithm.

8. A building energy management method applicable to multiple energy sources as described in claim 1, characterized in that, The SCADA system is an energy data acquisition and monitoring control system based on a strategy evolution mechanism.

9. A building energy management method applicable to multiple energy sources as described in claim 1, characterized in that, The scheduling strategy network based on energy efficiency assessment is a collaborative control network constructed based on deep deterministic strategy gradient.

10. A building energy management system suitable for multiple energy sources, used to execute a building energy management method suitable for multiple energy sources as described in any one of claims 1 to 9, characterized in that, The management system includes: The information coupling module is used to acquire multi-dimensional information flow of building energy based on the SCADA system, and then construct a multi-energy coupling map from the multi-dimensional information flow. The disturbance sensing module is used to perform distributed acquisition of the state data of each energy node in the multi-energy coupling spectrum and source-load disturbance modeling, thereby obtaining the energy disturbance set; The strategy reconstruction module is used to determine the adjustment tolerance range through the scheduling strategy network based on energy efficiency assessment and the energy disturbance set, and then reconstruct the strategy action of the multidimensional information flow through the adjustment tolerance range to obtain the attenuation constraint term. The fluctuation correction module is used to collect the fluctuation state of multidimensional energy flow inside the building, and to penalize and correct the fluctuation state according to the attenuation constraint term to obtain the offset calibration factor. The evolution update module is used to periodically evolve and update the output stream of the SCADA system based on the building's energy supply and demand data and the offset calibration factor, so as to obtain the building's energy optimization scheduling strategy stream.

Images