A graph neural network-based energy-saving method and system for central air-conditioning refrigeration rooms

Abstract

The present invention discloses a method and system for energy saving in a central air-conditioning and refrigeration room based on a graph neural network, comprising: S1, acquiring the operating data of each device in the central air-conditioning and refrigeration room, including operating status parameters, energy consumption data, and environmental parameters; S2, constructing a device topology diagram based on the collected data, defining the association relationship between devices in the form of nodes and edges, and dynamically updating the topology weights; S3, using the generated device topology diagram to train a graph neural network model, capturing the complex interaction relationship between devices, and generating an energy consumption feature vector; S4, combining real-time device operating data, using the trained model to predict energy consumption and generate a dynamic optimization strategy; S5, adjusting the operating parameters of the device according to the dynamic optimization strategy; S6, real-time monitoring of the adjusted device operating status, and feeding the optimized data back to the graph neural network model for updating. The present invention has the advantages of global optimization energy efficiency, precise data processing capabilities, and dynamic optimization response.

Description

Technical Field

[0001] The present invention relates to the field of energy control technology, and in particular to an energy-saving method and system for a central air-conditioning refrigeration room based on a graph neural network. Background Art

[0002] With the advancement of intelligent modern buildings, central air conditioning systems have become an indispensable component of large structures. However, as energy-intensive locations, energy optimization in central air conditioning and refrigeration rooms has become a key research focus in the field of energy conservation and emission reduction. Central air conditioning equipment operates under complex interdependencies and is influenced by multidimensional data, such as equipment operating parameters, energy consumption status, and environmental changes, increasing the complexity of system optimization.

[0003] Existing energy-saving methods for central air conditioning and refrigeration rooms typically rely on traditional optimization methods based on rules or statistical models. These methods exhibit the following significant shortcomings when faced with the dynamic changes in the operating environment and the complex connections between multiple devices:

[0004] 1. Difficulty in modeling complex device relationships: Traditional optimization methods cannot effectively capture the nonlinear and complex interactions between devices, which can easily lead to deviations in optimization results.

[0005] 2. Insufficient real-time prediction capabilities: Existing methods usually rely on static models in energy consumption prediction, which makes it difficult to adapt to changes in equipment operating status and environmental conditions in real time.

[0006] 3. Single optimization strategy: Traditional methods rely on fixed rules for optimization, lack dynamic adjustment capabilities, and are unable to fully utilize real-time data to generate targeted optimization solutions.

[0007] 4. Lack of feedback mechanism: Existing technologies generally lack closed-loop feedback control, and optimization results cannot be dynamically updated to the system, making it difficult to achieve continuous improvement and adaptive optimization.

[0008] Therefore, how to provide an energy-saving method and system for central air-conditioning refrigeration rooms based on graph neural networks is an urgent problem that technicians in this field need to solve. Summary of the Invention

[0009] One purpose of the present invention is to propose an energy-saving method for central air-conditioning refrigeration rooms based on graph neural networks. The present invention makes full use of the dynamic modeling and optimization capabilities of graph neural networks for complex relationships between devices, and describes in detail the steps for achieving real-time energy consumption prediction, dynamic optimization strategy generation and implementation. It has the advantages of high energy-saving efficiency, strong adaptability and good real-time performance.

[0010] According to an embodiment of the present invention, a method for energy saving in a central air-conditioning refrigeration room based on a graph neural network includes the following steps:

[0011] S1. Obtain multi-dimensional operating data of each device in the central air-conditioning and refrigeration room;

[0012] S2. Based on the collected equipment operation data, a topological diagram of the equipment in the central air-conditioning and refrigeration room is constructed. The association relationship between the equipment is defined as nodes and edges, and the topological weight is dynamically updated according to the association strength.

[0013] S3. Using the constructed device topology graph, we train a graph neural network model. By propagating and aggregating nodes layer by layer, we capture the nonlinear and complex interactions between devices and generate energy consumption feature vectors for each node.

[0014] S4: Based on the trained graph neural network model and combined with real-time equipment operation data, energy consumption is predicted and a dynamic optimization strategy for the overall energy consumption of the computer room is generated.

[0015] S5. Based on the generated dynamic optimization strategy for the overall energy consumption of the computer room, adjust the operating parameters of the equipment in the central air-conditioning and refrigeration room and coordinate the energy consumption distribution among the equipment;

[0016] S6. Monitor the equipment operation effect in real time, feed back the optimized operation data to the graph neural network model, and incrementally update the model parameters.

[0017] Optionally, the S2 specifically includes:

[0018] S21. Based on the physical layout and operating data characteristics of each device in the central air-conditioning and refrigeration room, define each device as a node. The node attributes include the device's real-time operating parameters, historical energy consumption data, and device type.

[0019] S22. Construct a set of edges based on the association relationship between devices. The topological weight of the edge is determined by calculating the interaction strength between devices:

[0020] ;

[0021] in, Representation device and equipment The topological weight between and Respectively represent devices and equipment In time The operating parameters of is the length of the time series;

[0022] S23. Based on the dynamic changes in the device operating status, the topological weight of the edge is updated in real time. The update rule determines whether the change in the interaction strength between devices exceeds the set range based on the preset interaction strength change threshold. If it exceeds the set range, the topological weight of the edge is recalculated;

[0023] S24. Generate a device topology graph based on the updated set of nodes and edges, and store the association relationship between the nodes and edges in a matrix form, where the value of the matrix element corresponds to the topological weight of the edge;

[0024] S25. Performing a sparse processing on the generated device topology graph, and removing edges whose association strength is lower than a set topology weight sparse threshold.

[0025] Optionally, the S3 specifically includes:

[0026] S31. Use the device topology map as input to initialize the node feature vectors of the graph neural network model, where the initial feature vector of each node is composed of the real-time operating parameters and historical energy consumption data of its device:

[0027] ;

[0028] in, For nodes The initial eigenvector of Representation node No. eigenvalues, is the dimension of node features;

[0029] S32, using a layer-by-layer propagation mechanism to update node features, and completing the propagation and update of feature vectors by aggregating neighbor node features and own features;

[0030] S33, iteratively executing the feature propagation and update process according to the set number of network layers and training objectives until the node feature vector aggregation is completed;

[0031] S34, performing linear transformation on the aggregated node feature vectors to generate an energy consumption feature vector for each node;

[0032] S35. The generated energy consumption feature vectors of all nodes are used as the output of the graph neural network.

[0033] Optionally, the S32 specifically includes:

[0034] S321. Determine the neighbor node set of each node based on the device topology map , where the nodes Representation device , Including nodes There are edges connecting all nodes;

[0035] S322, according to the topological weight , for each node Assign corresponding edge weights;

[0036] S323. Calculate the weighted sum of neighbor node features:

[0037] ;

[0038] in, For nodes In the The aggregation vector of the layer, Neighbor nodes In the The feature vector of the layer;

[0039] Merge Node Its own characteristics, combined with the trainable weight matrix , computing nodes Updated eigenvector of :

[0040] ;

[0041] in, is a nonlinear activation function, Representation node In the The feature vector of the layer;

[0042] S324. Repeat the feature aggregation and update process in S323 according to the set number of propagation layers until the propagation of the specified number of layers is completed and the final node feature vector is generated.

[0043] Optionally, the S34 specifically includes:

[0044] S341, the final node feature vector As input, where Representation node Feature vector in the last layer of the graph neural network;

[0045] S342. Set the linear transformation weight matrix ,in The dimension is , express Dimensions, Represents the target dimension for generating energy consumption feature vector;

[0046] S343, for each node The eigenvector of Apply linear transformation to generate the energy consumption feature vector of the node:

[0047] ;

[0048] in, Representation node The energy consumption characteristic vector of is the bias vector of the linear transformation, with the same dimension as same.

[0049] Optionally, the S4 specifically includes:

[0050] S41. Inputting real-time device operation data into the trained graph neural network model, where the device operation data includes the device's real-time energy consumption, operating status, and environmental parameters;

[0051] S42. Use the graph neural network model to predict energy consumption based on the input real-time equipment operation data:

[0052] ;

[0053] in, Indicates the predicted overall energy consumption of the computer room, and are the trainable weight matrices for the first and second layers respectively, and is the corresponding bias vector, is a nonlinear activation function, is the device node feature matrix, is the adjacency matrix of the device topology graph;

[0054] S43. Based on the predicted overall energy consumption of the computer room , combined with the energy consumption optimization goal, calculate the dynamic optimization strategy;

[0055] S44. Generate an equipment operation adjustment plan based on the calculated dynamic optimization strategy, including equipment start and stop sequence, operating load, and parameter control values.

[0056] Optionally, the S43 specifically includes:

[0057] S431, based on the predicted overall energy consumption of the computer room , combined with the energy consumption optimization goal, define the energy consumption cost function ,in Represents a set of device operating parameters Corresponding energy consumption:

[0058] ;

[0059] in, Indicates the The power consumption of each device, Indicates the The running time of each device, is the total number of devices in the computer room;

[0060] S432. Setting constraint condition function , including equipment operating load limits, start and stop conditions and environmental constraints:

[0061] ;

[0062] in, Indicates the The maximum permissible load of a device, Used to limit the operating power of the equipment to not exceed the permitted range;

[0063] S433. Calculate the device operation parameter set through the optimization function , the optimization goal is to minimize the energy cost function and constraint function :

[0064] ;

[0065] in, is the trade-off coefficient used to balance energy consumption cost and constraints;

[0066] S434, using iterative optimization algorithm to solve the optimization target, by gradually adjusting the equipment operating parameter set , until the set convergence condition or iteration limit is reached, generating a dynamic optimization strategy .

[0067] Optionally, the S5 specifically includes:

[0068] S51. According to the dynamic optimization strategy , parse the device operating parameter set ;

[0069] S52, based on the optimization parameter set , adjust the operating status of each device, the device start and stop status variables are :

[0070] ;

[0071] S53, according to the power control value And load distribution requirements, adjust the operating power of the running equipment:

[0072] ;

[0073] in, Indicates the adjusted device The operating power, and Respectively represent devices Maximum and minimum permissible operating power;

[0074] S54, real-time monitoring of the adjusted equipment operating status, and feedback of the equipment operating parameter adjustment values ​​to the central control system, so that the power distribution of each device meets the dynamic optimization strategy requirements.

[0075] A central air-conditioning refrigeration room energy-saving system based on graph neural network, including:

[0076] Data acquisition device, used to obtain multi-dimensional operating data of each device in the central air-conditioning and refrigeration room, including the real-time operating parameters of the equipment, historical energy consumption data and environmental parameters;

[0077] A graph structure generating device for constructing a device topology graph of a central air-conditioning refrigeration room based on the data acquired by the data acquisition device, defining the association relationships between devices in the form of nodes and edges, and dynamically updating the topology weights according to the association strength;

[0078] A graph neural network model training device is used to train the graph neural network model using the generated device topology graph. By propagating and aggregating node features layer by layer, it captures the nonlinear and complex interaction relationships between devices and generates an energy consumption feature vector for each node.

[0079] Energy consumption prediction and optimization device, which combines real-time equipment operation data with a trained graph neural network model to predict energy consumption and generate a dynamic optimization strategy for the overall energy consumption of the central air conditioning and refrigeration room;

[0080] An operating parameter adjustment device is used to adjust the operating parameters of the equipment in the central air-conditioning refrigeration room according to the generated dynamic optimization strategy, including the equipment start and stop status, operating power and load distribution;

[0081] The feedback control device is used to monitor the operating status of the adjusted equipment in real time and feed back the adjusted values ​​of the operating parameters to the system to achieve dynamic optimization of energy consumption distribution among the equipment.

[0082] The beneficial effects of the present invention are:

[0083] (1) By combining the complex relationship modeling capabilities of graph neural networks and dynamic optimization algorithms, the present invention provides a deep understanding and dynamic adaptability of the complex energy consumption relationships between devices in central air-conditioning and refrigeration rooms, enabling the system to predict energy consumption in real time and generate optimization strategies, thereby effectively reducing overall energy consumption, especially in energy-saving optimization in scenarios with nonlinear interactions among multiple devices.

[0084] (2) This invention achieves dynamic management and adaptive regulation of energy consumption in central air-conditioning and refrigeration rooms through real-time data acquisition, equipment status monitoring, and closed-loop feedback control of optimization strategies. This not only improves the real-time nature of energy-saving management but also enhances the reliability and effectiveness of the optimization strategies under different operating conditions.

[0085] (3) The present invention provides a comprehensive solution for energy-saving management of central air-conditioning and refrigeration rooms by combining multi-dimensional operation data analysis, energy consumption prediction and optimization, and dynamic adjustment of equipment operation parameters. This enables the system to continuously learn and adapt to changes in the operating environment, thereby achieving intelligent, precise and efficient energy-saving management. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0087] Figure 1 This is the overall framework diagram of the energy-saving method and system for central air-conditioning refrigeration rooms based on graph neural networks proposed in the present invention. DETAILED DESCRIPTION

[0088] The present invention will now be described in further detail with reference to the accompanying drawings, which are simplified schematic diagrams that illustrate the basic structure of the present invention in a schematic manner.

[0089] refer to Figure 1 , a central air-conditioning refrigeration room energy-saving method based on graph neural network, comprising the following steps:

[0090] S1. Obtain multi-dimensional operating data of each device in the central air-conditioning and refrigeration room;

[0091] S2. Based on the collected equipment operation data, a topological diagram of the equipment in the central air-conditioning and refrigeration room is constructed. The association relationship between the equipment is defined as nodes and edges, and the topological weight is dynamically updated according to the association strength.

[0092] S3. Using the constructed device topology graph, we train a graph neural network model. By propagating and aggregating nodes layer by layer, we capture the nonlinear and complex interactions between devices and generate energy consumption feature vectors for each node.

[0093] S4: Based on the trained graph neural network model and combined with real-time equipment operation data, energy consumption is predicted and a dynamic optimization strategy for the overall energy consumption of the computer room is generated.

[0094] S5. Based on the generated dynamic optimization strategy for the overall energy consumption of the computer room, adjust the operating parameters of the equipment in the central air-conditioning and refrigeration room and coordinate the energy consumption distribution among the equipment;

[0095] S6. Monitor the equipment operation effect in real time, feed back the optimized operation data to the graph neural network model, and incrementally update the model parameters.

[0096] In this embodiment, S2 specifically includes:

[0097] S21. Based on the physical layout and operating data characteristics of each device in the central air-conditioning and refrigeration room, define each device as a node. The node attributes include the device's real-time operating parameters, historical energy consumption data, and device type.

[0098] S22. Construct a set of edges based on the association relationship between devices. The topological weight of the edge is determined by calculating the interaction strength between devices:

[0099] ;

[0100] in, Representation device and equipment The topological weight between and Respectively represent devices and equipment In time The operating parameters of is the length of the time series;

[0101] S23. Based on the dynamic changes in the device operating status, the topological weight of the edge is updated in real time. The update rule determines whether the change in the interaction strength between devices exceeds the set range based on the preset interaction strength change threshold. If it exceeds the set range, the topological weight of the edge is recalculated;

[0102] S24. Generate a device topology graph based on the updated set of nodes and edges, and store the association between the nodes and edges in a matrix format, where the values ​​of the matrix elements correspond to the topological weights of the edges;

[0103] S25. Performing a sparse processing on the generated device topology graph, and removing edges whose association strength is lower than a set topology weight sparse threshold.

[0104] In this embodiment, S3 specifically includes:

[0105] S31. Use the device topology map as input to initialize the node feature vectors of the graph neural network model, where the initial feature vector of each node is composed of the real-time operating parameters and historical energy consumption data of its device:

[0106] ;

[0107] in, For nodes The initial eigenvector of Representation node No. eigenvalues, is the dimension of node features;

[0108] S32, using a layer-by-layer propagation mechanism to update node features, and completing the propagation and update of feature vectors by aggregating neighbor node features and own features;

[0109] S33, iteratively executing the feature propagation and update process according to the set number of network layers and training objectives until the node feature vector aggregation is completed;

[0110] S34, performing linear transformation on the aggregated node feature vectors to generate an energy consumption feature vector for each node;

[0111] S35. The generated energy consumption feature vectors of all nodes are used as the output of the graph neural network.

[0112] In this embodiment, S32 specifically includes:

[0113] S321. Determine the neighbor node set of each node based on the device topology map , where the nodes Representation device , Including nodes There are edges connecting all nodes;

[0114] S322, according to the topological weight , for each node Assign corresponding edge weights;

[0115] S323. Calculate the weighted sum of neighbor node features:

[0116] ;

[0117] in, For nodes In the The aggregation vector of the layer, Neighbor nodes In the The feature vector of the layer;

[0118] Merge Node Its own characteristics, combined with the trainable weight matrix , computing nodes Updated eigenvector of :

[0119] ;

[0120] in, is a nonlinear activation function, Representation node In the The feature vector of the layer;

[0121] S324. Repeat the feature aggregation and update process in S323 according to the set number of propagation layers until the propagation of the specified number of layers is completed and the final node feature vector is generated.

[0122] In this embodiment, S34 specifically includes:

[0123] S341, the final node feature vector As input, where Representation node Feature vector in the last layer of the graph neural network;

[0124] S342. Set the linear transformation weight matrix ,in The dimension is , express Dimensions, Represents the target dimension for generating energy consumption feature vector;

[0125] S343, for each node The eigenvector of Apply linear transformation to generate the energy consumption feature vector of the node:

[0126] ;

[0127] in, Representation node The energy consumption characteristic vector of is the bias vector of the linear transformation, with the same dimension as same.

[0128] In this embodiment, S4 specifically includes:

[0129] S41. Inputting real-time device operation data into the trained graph neural network model, where the device operation data includes the device's real-time energy consumption, operating status, and environmental parameters;

[0130] S42. Use the graph neural network model to predict energy consumption based on the input real-time equipment operation data:

[0131] ;

[0132] in, Indicates the predicted overall energy consumption of the computer room, and are the trainable weight matrices for the first and second layers respectively, and is the corresponding bias vector, is a nonlinear activation function, is the device node feature matrix, is the adjacency matrix of the device topology graph;

[0133] S43. Based on the predicted overall energy consumption of the computer room , combined with the energy consumption optimization goal, calculate the dynamic optimization strategy;

[0134] S44. Generate an equipment operation adjustment plan based on the calculated dynamic optimization strategy, including equipment start and stop sequence, operating load, and parameter control values.

[0135] In this embodiment, S43 specifically includes:

[0136] S431, based on the predicted overall energy consumption of the computer room , combined with the energy consumption optimization goal, define the energy consumption cost function ,in Represents a set of device operating parameters Corresponding energy consumption:

[0137] ;

[0138] in, Indicates the The power consumption of each device, Indicates the The running time of each device, is the total number of devices in the computer room;

[0139] S432. Setting constraint condition function , including equipment operating load limits, start and stop conditions and environmental constraints:

[0140] ;

[0141] in, Indicates the The maximum permissible load of a device, Used to limit the operating power of the equipment to not exceed the permitted range;

[0142] S433. Calculate the device operation parameter set through the optimization function , the optimization goal is to minimize the energy cost function and constraint function :

[0143] ;

[0144] in, is the trade-off coefficient used to balance energy consumption cost and constraints;

[0145] In this embodiment, S5 specifically includes:

[0146] S51. According to the dynamic optimization strategy , parse the device operating parameter set ;

[0147] S52, based on the optimization parameter set , adjust the operating status of each device, the device start and stop status variables are :

[0148] ;

[0149] S53, according to the power control value And load distribution requirements, adjust the operating power of the running equipment:

[0150] ;

[0151] in, Indicates the adjusted device The operating power, and Respectively represent devices Maximum and minimum permissible operating power;

[0152] S54, real-time monitoring of the adjusted equipment operating status, and feedback of the equipment operating parameter adjustment values ​​to the central control system, so that the power distribution of each device meets the dynamic optimization strategy requirements.

[0153] A central air-conditioning refrigeration room energy-saving system based on graph neural network, including:

[0154] The operating parameter control module monitors the operating status of refrigeration equipment in real time based on a multi-dimensional energy consumption control strategy, obtains the equipment's energy consumption and environmental data, and calculates the optimal operating parameters for each device based on a graph neural network model;

[0155] The start-stop control module adjusts the start-stop sequence of the equipment based on the energy consumption forecast results and real-time data of the equipment. The adjustment of the start-stop sequence is calculated and optimized by the operation parameter control module;

[0156] The compressor frequency control module adjusts the operating frequency of the compressor according to the energy consumption status and load requirements of the equipment;

[0157] The refrigerant flow control module calculates and optimizes the refrigerant flow distribution plan based on the real-time cooling load demand and energy efficiency requirements among equipment;

[0158] The fan control module calculates the optimal operating speed of the fan based on the current energy efficiency requirements of the equipment and environmental conditions;

[0159] The feedback data collection module monitors the actual energy consumption data and environmental variables of the equipment in real time, and inputs the feedback data into the graph neural network model for updating and optimization;

[0160] The graph neural network module updates node feature vectors and edge weight parameters based on real-time feedback data and global energy consumption prediction results, and continuously optimizes the graph neural network model to generate a globally optimal energy-saving operation strategy;

[0161] The energy efficiency control strategy generation module generates a dynamic energy-saving control strategy that adapts to environmental changes and load demands based on the updated node features and edge weights after the graph neural network model converges.

[0162] Example 1:

[0163] In order to verify the feasibility of the present invention, we applied the present invention to the energy-saving management system of a central air-conditioning and refrigeration room in a large commercial building. The building consists of multiple floors and different functional areas. The energy consumption of the air-conditioning and refrigeration system accounts for a large proportion of the overall energy consumption, especially in summer and during high-load usage. Due to the uncertain operational needs of the building, traditional energy-saving methods rely on fixed control strategies and cannot cope with real-time changes in the environment and load demands, resulting in unnecessary energy waste in the system during peak load periods. In order to address this problem, the building management decided to adopt the energy-saving method for central air-conditioning and refrigeration rooms based on graph neural networks proposed in this invention, so as to improve the energy efficiency of the system and reduce unnecessary energy consumption.

[0164] In this application scenario, the graph neural network model of the present invention first monitors the operating status of the refrigeration equipment in real time, collects the energy consumption data and environmental data of each device, and constructs a dynamic optimization model of energy efficiency interaction between devices by updating and optimizing the node feature vectors and edge weights. With the change of seasons and fluctuations in external temperature, the load demand of the refrigeration room has changed significantly. Traditional control methods are often unable to adjust in time, resulting in energy waste during high load periods. The present invention automatically adjusts the start and stop sequence of the equipment, compressor frequency, refrigerant flow distribution and fan speed by dynamically learning and adjusting the collaborative relationship between devices, so that the system can respond quickly when the load changes, thereby achieving global optimal energy efficiency.

[0165] During the implementation process, the system continuously updates the energy efficiency control strategy of the equipment through the energy consumption data and environmental variables fed back by the real-time data acquisition module. Through the training and optimization of the graph neural network model, the node feature vectors and edge weights are continuously adjusted, thereby optimizing the energy efficiency interaction and collaborative work between devices. The nonlinear relationship and real-time changes between devices are often not fully considered in traditional methods, but the present invention successfully overcomes this problem through dynamic learning based on graph neural networks. As the system is optimized, the graph neural network can adjust the operating strategy of the equipment according to the feedback data, especially in the event of load fluctuations and environmental changes, and can respond in a timely manner.

[0166] This invention also incorporates a feedback mechanism, enabling the system to continuously adjust itself through real-time monitoring and data analysis. Whenever the system detects a discrepancy between a device's actual energy consumption and its predicted results, it immediately initiates a model update to optimize the energy efficiency interactions between devices. This adaptive adjustment mechanism allows the system to continuously improve energy efficiency over the long term, avoiding the constraints of manual intervention and fixed rules, resulting in sustained improvements in energy savings.

[0167] To evaluate the effectiveness of this invention, we implemented the method in the building and compared the system's energy efficiency before and after implementation. The following table shows the energy savings in the building's central air conditioning and refrigeration room before and after the system was deployed:

[0168]

[0169] As can be seen from the above data table, the system after the application of the present invention has significantly improved energy efficiency. Especially during high-load operation, the traditional system cannot flexibly adjust the collaborative relationship between equipment, resulting in excessive energy consumption. The present invention greatly improves the energy saving effect through real-time monitoring and dynamic optimization. For example, the total monthly energy consumption was reduced from 52,000 kWh before deployment to 39,000 kWh, and the energy saving rate reached 25%. In addition, the optimization adjustment of parameters such as equipment start-stop sequence, compressor frequency, refrigerant flow distribution and fan speed also demonstrates the system's efficient adaptability in a dynamically changing environment.

[0170] Furthermore, after implementing this invention, the system can continuously improve energy efficiency through real-time data feedback and adaptive optimization mechanisms. During specific time periods (such as peak summer months or extreme external temperature fluctuations), the system can quickly identify load changes and automatically adjust the equipment's operating mode, ensuring the refrigeration system always maintains optimal energy efficiency. This not only reduces unnecessary energy consumption but also significantly extends the equipment's lifespan and minimizes equipment wear and tear caused by excessive operation.

[0171] As demonstrated in the aforementioned examples, the graph neural network-based energy-saving method for central air-conditioning and refrigeration rooms of the present invention effectively addresses the inability of traditional energy-saving methods to adapt to environmental changes and load fluctuations. It optimizes the coordinated operation of equipment in real time during dynamic operation, significantly improving the overall energy efficiency of the system and achieving the desired energy-saving effects. Through efficient data processing, dynamic optimization, and adaptive adjustment, the system ensures the long-term sustainability of energy-saving effects while reducing reliance on manual intervention and fixed control strategies, further enhancing the intelligence and efficiency of energy-saving management.

[0172] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A central air conditioning refrigeration room energy-saving method based on graph neural network, characterized in that: The steps include: S1. Obtain multi-dimensional operating data of each device in the central air-conditioning and refrigeration room; S2. Based on the collected equipment operation data, a topological diagram of the equipment in the central air-conditioning and refrigeration room is constructed. The association relationship between the equipment is defined as nodes and edges, and the topological weight is dynamically updated according to the association strength. The S2 specifically includes: S21. Based on the physical layout and operating data characteristics of each device in the central air-conditioning and refrigeration room, define each device as a node. The node attributes include the device's real-time operating parameters, historical energy consumption data, and device type. S22. Construct a set of edges based on the association relationship between devices. The topological weight of the edge is determined by calculating the interaction strength between devices: Among them, w ij represents the topological weight between device i and device j, x i (t) and x j (t) represents the operating parameters of device i and device j at time t, respectively, and T is the length of the time series; S23. Based on the dynamic changes in the device operating status, the topological weight of the edge is updated in real time. The update rule determines whether the change in the interaction strength between devices exceeds the set range based on the preset interaction strength change threshold. If it exceeds the set range, the topological weight of the edge is recalculated; S24. Generate a device topology graph based on the updated set of nodes and edges, and store the association between the nodes and edges in a matrix format, where the values ​​of the matrix elements correspond to the topological weights of the edges; S25. Performing a sparse processing on the generated device topology graph, removing edges whose association strength is lower than a set topology weight sparse threshold; S3. Using the constructed device topology graph, we train a graph neural network model. By propagating and aggregating nodes layer by layer, we capture the nonlinear and complex interactions between devices and generate energy consumption feature vectors for each node. The S3 specifically includes: S31. Use the device topology map as input to initialize the node feature vectors of the graph neural network model, where the initial feature vector of each node is composed of the real-time operating parameters and historical energy consumption data of its device: in, is the initial eigenvector of node i, x i,k represents the kth eigenvalue of node i, and m is the dimension of the node feature; S32, using a layer-by-layer propagation mechanism to update node features, and completing the propagation and update of feature vectors by aggregating neighbor node features and own features; S33, iteratively executing the feature propagation and update process according to the set number of network layers and training objectives until the node feature vector aggregation is completed; S34, performing linear transformation on the aggregated node feature vectors to generate an energy consumption feature vector for each node; S35. The generated energy consumption feature vectors of all nodes are used as the output of the graph neural network; S4: Based on the trained graph neural network model and combined with real-time equipment operation data, energy consumption is predicted and a dynamic optimization strategy for the overall energy consumption of the computer room is generated. The S4 specifically includes: S41. Inputting real-time device operation data into the trained graph neural network model, where the device operation data includes the device's real-time energy consumption, operating status, and environmental parameters; S42. Use the graph neural network model to predict energy consumption based on the input real-time equipment operation data: in, represents the predicted overall energy consumption of the computer room, W1 and W2 are the trainable weight matrices of the first and second layers respectively, b1 and b2 are the corresponding bias vectors, ReLU is the nonlinear activation function, H is the device node feature matrix, and A is the adjacency matrix of the device topology graph; S43. Based on the predicted overall energy consumption of the computer room Combined with the energy consumption optimization goal, calculate the dynamic optimization strategy; S44. Generate an equipment operation adjustment plan based on the calculated dynamic optimization strategy, including equipment start-up and shutdown sequence, operating load, and parameter control values; S5. Based on the generated dynamic optimization strategy for the overall energy consumption of the computer room, adjust the operating parameters of the equipment in the central air-conditioning and refrigeration room and coordinate the energy consumption distribution among the equipment; The S5 specifically includes: S51. Analyze the device operating parameter set P according to the dynamic optimization strategy O; S52, according to the optimized parameter set P, adjust the operating state of each device, and the device start and stop state variable is S i : S53, according to the power control value P i Adjust the operating power of the running equipment according to the load distribution requirements: P i new =min(P i max ,max(P i min ,P i )); Among them, P i new represents the adjusted operating power of device i, P i max and P i min Respectively represent the maximum and minimum allowed operating power of device i; S54, real-time monitoring of the adjusted equipment operating status, and feeding back the equipment operating parameter adjustment values ​​to the central control system so that the power allocation of each device meets the requirements of the dynamic optimization strategy O; S6. Monitor the equipment operation effect in real time, feed back the optimized operation data to the graph neural network model, and incrementally update the model parameters.

2. The energy-saving method for central air-conditioning refrigeration room based on graph neural network according to claim 1 is characterized in that: The S32 specifically includes: S321. Determine the neighbor node set N(i) of each node based on the device topology graph, where node i represents device i and N(i) includes all nodes that have edges connected to node i. S322, according to the topological weight w ij , assign corresponding edge weight to each node j∈N(i); S323. Calculate the weighted sum of neighbor node features: in, is the aggregation vector of node i at layer l, is the feature vector of neighbor node j in layer l; Merge the characteristics of node i and combine them with the trainable weight matrix W (l) , calculate the updated feature vector of node i: Among them, σ is a nonlinear activation function, Represents the feature vector of node i in layer l+1; S324. Repeat the feature aggregation and update process in S323 according to the set number of propagation layers until the propagation of the specified number of layers is completed and the final node feature vector is generated.

3. The energy-saving method for central air-conditioning refrigeration room based on graph neural network according to claim 1 is characterized in that: The S34 specifically includes: S341, the final node feature vector As input, where Represents the feature vector of node i in the last layer of the graph neural network; S342, set the linear transformation weight matrix W e , where W e The dimension is (d L ×d E ), d L express The dimension, d E Represents the target dimension for generating energy consumption feature vector; S343, the feature vector of each node i Apply linear transformation to generate the energy consumption feature vector of the node: Among them, e i represents the energy consumption feature vector of node i, b e is the bias vector of the linear transformation, with the same dimension as e i same.

4. The energy-saving method for central air-conditioning refrigeration room based on graph neural network according to claim 1 is characterized in that: The S43 specifically includes: S431, based on the predicted overall energy consumption of the computer room Combined with the energy consumption optimization goal, the energy consumption cost function C(P) is defined, where C(P) represents the energy consumption corresponding to the equipment operating parameter set P: Among them, P i represents the power consumption of the i-th device, T i represents the operating time of the i-th device, and N is the total number of devices in the computer room; S432. Set the constraint function R(P), including equipment load limit, start and stop conditions, and environmental constraints: Among them, L i Indicates the maximum allowable load of the i-th device, max(0,L i -P i ) is used to limit the operating power of the equipment to not exceed the permitted range; S433. Calculate the device operating parameter set P through the optimization function. The optimization goal is to minimize the energy cost function C(P) and the constraint condition function R(P) at the same time: O=a P (C(P)+λ·R(P)): Among them, λ is the trade-off coefficient, which is used to balance the energy cost and the constraints; S434. An iterative optimization algorithm is used to solve the optimization target, and a dynamic optimization strategy O is generated by gradually adjusting the device operating parameter set P until the set convergence condition or the iteration number limit is reached.

5. The central air conditioning refrigeration room energy saving method based on graph neural network and the central air conditioning refrigeration room energy saving system based on graph neural network according to any one of claims 1 to 4 are characterized in that: include: Data acquisition device, used to obtain multi-dimensional operating data of each device in the central air-conditioning and refrigeration room, including the real-time operating parameters of the equipment, historical energy consumption data and environmental parameters; A graph structure generating device for constructing a device topology graph of a central air-conditioning refrigeration room based on the data acquired by the data acquisition device, defining the association relationships between devices in the form of nodes and edges, and dynamically updating the topology weights according to the association strength; A graph neural network model training device is used to train the graph neural network model using the generated device topology graph. By propagating and aggregating node features layer by layer, it captures the nonlinear and complex interaction relationships between devices and generates an energy consumption feature vector for each node. Energy consumption prediction and optimization device, which combines real-time equipment operation data with a trained graph neural network model to predict energy consumption and generate a dynamic optimization strategy for the overall energy consumption of the central air conditioning and refrigeration room; An operating parameter adjustment device is used to adjust the operating parameters of the equipment in the central air-conditioning refrigeration room according to the generated dynamic optimization strategy, including the equipment start and stop status, operating power and load distribution; The feedback control device is used to monitor the operating status of the adjusted equipment in real time and feed back the adjusted values ​​of the operating parameters to the system to achieve dynamic optimization of energy consumption distribution among the equipment.

Images