Multi-connected air conditioner group cooperative scheduling method

Abstract

The application relates to the technical field of intelligent building air conditioner control, and in particular to a multi-connected air conditioner group cooperative scheduling method. The method determines the characteristics and coupling relationships of each indoor unit node based on air conditioner operation, space layout and system topology data; maps the indoor units as graph nodes, constructs multi-type coupling edges according to the coupling relationships of refrigerant systems, horizontal spaces and vertical spaces, forms a group topology graph, dynamically updates the edge weights based on the topology graph and node characteristics, generates a dynamic adjacency matrix, extracts space coupling characteristics and multi-scale time characteristics, performs gating fusion on the two types of characteristics to obtain spatiotemporal fusion characteristics, performs hierarchical scheduling decision to generate candidate scheduling data, and finally obtains a target cooperative scheduling strategy through physical constraint processing to control the operation of the multi-connected air conditioner group.

Description

Technical Field

[0001] This application relates to the field of intelligent control technology for building air conditioning, and in particular to a method for collaborative scheduling of multi-split air conditioning groups. Background Technology

[0002] Multi-split air conditioning systems are widely used in large buildings such as commercial buildings, hospitals, campuses, and hotels. They typically consist of one or more outdoor units and multiple indoor units connected to them. The indoor units are distributed across different rooms, floors, or functional areas and are linked to their corresponding outdoor units via refrigerant loops. In actual operation, multi-split air conditioning systems are affected not only by changes in the indoor and outdoor environment, occupant activity, and room load, but also by the refrigerant distribution within the same outdoor unit system and the heat transfer between adjacent spaces. Existing centralized building air conditioning control systems typically employ rule-based control methods, such as controlling on / off according to a fixed schedule, controlling the operation of a single air conditioner according to indoor temperature thresholds, or rotating operations according to preset groups. This type of control primarily focuses on independent control of individual air conditioners or fixed groups, failing to adequately consider the refrigerant system coupling relationships between different indoor units, the horizontal spatial heat coupling relationships between adjacent areas on the same floor, and the vertical spatial heat coupling relationships between floors. Therefore, when multiple areas experience simultaneous changes in cooling and heating loads, problems such as independent control strategies for different indoor air conditioning units, repeated cooling or heating in adjacent areas, and uncoordinated load distribution in outdoor unit systems can easily arise, leading to poor centralized control performance of building air conditioning systems. Thus, improving the synergy of centralized control of building air conditioning systems has become an urgent technical problem to be solved. Summary of the Invention

[0003] The main purpose of this application is to provide a method for coordinated scheduling of multi-split air conditioning groups, which aims to solve the technical problem of how to improve the coordination of centralized control of building air conditioning.

[0004] To achieve the above objectives, this application provides a method for coordinated scheduling of multi-split air conditioning groups, the method comprising the following steps: Based on air conditioning operation data, spatial layout data, and system topology data, the node characteristic data corresponding to each air conditioning indoor unit and the coupling relationship data between air conditioning indoor units are determined. Based on the node feature data and the coupling relationship data, each indoor air conditioner unit is taken as a graph node, and corresponding multi-type coupling edges are constructed according to the refrigerant system coupling relationship, horizontal spatial coupling relationship and vertical spatial coupling relationship in the coupling relationship data to obtain the topology graph of the multi-split air conditioner group. Based on the topology of the multi-split air conditioning group and the node feature data, the edge weights of each type of coupling edge are dynamically updated to obtain a dynamic adjacency matrix. Based on the dynamic adjacency matrix and the node feature data, graph attention spatial feature extraction and multi-scale temporal feature extraction are performed to obtain spatial coupling feature data and temporal operation feature data. The spatial coupling feature data and the temporal operation feature data are subjected to gated fusion processing to obtain spatiotemporal fusion feature data, and hierarchical scheduling decision processing is performed based on the spatiotemporal fusion feature data to obtain candidate collaborative scheduling data. The candidate coordinated scheduling data is constrained based on preset physical constraints to obtain a target coordinated scheduling strategy, and the multi-split air conditioning group is scheduled and controlled according to the target coordinated scheduling strategy.

[0005] In one embodiment, the step of determining the node characteristic data corresponding to each indoor air conditioner unit and the coupling relationship data between the indoor air conditioners based on air conditioner operation data, spatial layout data, and system topology data includes: The air conditioner operation data is analyzed to determine the temperature status data, operation status data and energy consumption status data of each indoor unit. Based on the temperature status data, operation status data and energy consumption status data, the corresponding operation characteristic data of each indoor unit is generated. The system topology data is parsed to determine the outdoor unit affiliation relationship, the outdoor unit system to which each indoor air conditioner unit belongs and the refrigerant association information of each indoor air conditioner unit in the corresponding outdoor unit system, and the refrigerant system coupling relationship between indoor air conditioner units is determined based on the outdoor unit system and the refrigerant association information. Based on the spatial layout data, the floor position relationship and spatial adjacency relationship of each air conditioner indoor unit are determined, and the horizontal spatial coupling relationship and vertical spatial coupling relationship between the air conditioner indoor units are determined according to the floor position relationship and the spatial adjacency relationship. The operating characteristic data is associated with the corresponding air conditioner indoor unit to obtain the node characteristic data corresponding to each air conditioner indoor unit. The coupling relationship data between the air conditioner indoor units is obtained according to the refrigerant system coupling relationship, the horizontal spatial coupling relationship and the vertical spatial coupling relationship.

[0006] In one embodiment, the step of using each indoor air conditioning unit as a graph node based on the node feature data and the coupling relationship data, and constructing corresponding multi-type coupling edges according to the refrigerant system coupling relationship, horizontal spatial coupling relationship, and vertical spatial coupling relationship in the coupling relationship data to obtain the topology graph of the multi-split air conditioning group includes: Based on the node feature data, node mapping processing is performed on each indoor air conditioner unit to determine the graph nodes that correspond one-to-one with each indoor air conditioner unit, and each graph node is associated with the corresponding node feature data to obtain a set of indoor air conditioner unit nodes. Based on the refrigerant system coupling relationship in the coupling relationship data, refrigerant association matching is performed on the graph nodes belonging to the same outdoor unit system in the set of indoor air conditioning nodes to construct refrigerant system coupling edges. Based on the horizontal spatial coupling relationship and the vertical spatial coupling relationship, horizontal spatial coupling edges and vertical spatial coupling edges are constructed respectively for the graph nodes with spatial thermal coupling relationship in the set of indoor air conditioning nodes. The refrigerant system coupling edges, the horizontal space coupling edges, and the vertical space coupling edges are labeled with edge types and their node connection relationships are integrated to obtain a set of multiple coupling edges corresponding to the set of indoor air conditioning unit nodes. Based on the set of indoor air conditioning unit nodes and the set of multiple coupling edges, the topology diagram of the multi-split air conditioning group is generated.

[0007] In one embodiment, the step of dynamically updating the edge weights of various types of coupling edges based on the topology graph of the multi-split air conditioning group and the node feature data to obtain a dynamic adjacency matrix, and performing graph attention spatial feature extraction and multi-scale temporal feature extraction based on the dynamic adjacency matrix and the node feature data to obtain spatial coupling feature data and temporal operation feature data includes: Based on the edge type information, node connection relationship, and temperature status information in the node feature data of each multi-type coupling edge in the topology graph of the multi-split air conditioning group, the coupling strength corresponding to each multi-type coupling edge is dynamically characterized to obtain the updated edge weight corresponding to each multi-type coupling edge, and the dynamic adjacency matrix is ​​generated according to the updated edge weight. Based on the dynamic adjacency matrix, the set of neighboring nodes corresponding to each graph node is determined. Then, according to the node feature data, the edge type information, and the updated edge weight, graph attention calculation and feature aggregation processing are performed on the spatial influence weight between each graph node and its corresponding neighboring node to obtain the spatial coupling feature data corresponding to each air conditioner indoor unit. Based on the node feature data, a historical node feature sequence corresponding to each air conditioner indoor unit is constructed, and multi-scale causal convolution processing and scale fusion processing are performed on the historical node feature sequence to obtain the time operation feature data corresponding to each air conditioner indoor unit.

[0008] In one embodiment, the step of performing gated fusion processing on the spatial coupling feature data and the temporal runtime feature data to obtain spatiotemporal fusion feature data, and performing hierarchical scheduling decision processing based on the spatiotemporal fusion feature data to obtain candidate cooperative scheduling data includes: The spatial coupling feature data and the temporal operation feature data are subjected to feature dimension matching and association encoding to obtain spatiotemporal candidate representation data, and the gating weight data corresponding to each air conditioner indoor unit is determined based on the spatiotemporal candidate representation data. Based on the gating weight data, the spatial coupling feature data and the temporal operation feature data are weighted and fused to obtain the spatiotemporal fusion feature data corresponding to each air conditioner indoor unit. According to the affiliation relationship between the air conditioner indoor unit and the outdoor unit system, the spatiotemporal fusion feature data is aggregated at the outdoor unit system level to obtain the outdoor unit system aggregated feature data. Global load allocation decisions are made based on the aggregated feature data of the outdoor system to obtain outdoor system load allocation data. In addition, indoor parameter adjustment decisions and timing execution planning are made based on the outdoor system load allocation data and the spatiotemporal fusion feature data to obtain candidate collaborative scheduling data.

[0009] In one embodiment, the step of constraining the candidate coordinated scheduling data based on preset physical constraints to obtain a target coordinated scheduling strategy, and then performing scheduling control on the multi-split air conditioning group according to the target coordinated scheduling strategy, includes: The candidate collaborative scheduling data is parsed to determine the load distribution data of the external system, the parameter adjustment data of the internal system, and the time-series execution plan data. Based on the preset physical constraints, constraint verification objects are established for the load distribution data of the external system, the parameter adjustment data of the internal system, and the time-series execution plan data, respectively. Based on the constraint verification object, the feasibility of the candidate collaborative scheduling data is verified by outdoor unit capacity constraints, temperature change rate constraints, minimum running time constraints, and refrigerant balance constraints. The candidate collaborative scheduling data that does not meet the preset physical constraint conditions are subjected to constraint correction processing to obtain constrained collaborative scheduling data. Based on the constrained collaborative scheduling data, a target collaborative scheduling strategy is generated, and load allocation instructions, parameter adjustment instructions, and timing execution instructions are issued to the corresponding outdoor unit system and indoor air conditioning unit according to the target collaborative scheduling strategy, so as to perform scheduling control on the multi-split air conditioning group.

[0010] Furthermore, to achieve the above objectives, this application also proposes a multi-split air conditioning group collaborative scheduling device, which includes: The coupling relationship module is used to determine the node characteristic data corresponding to each air conditioner indoor unit and the coupling relationship data between the air conditioner indoor units based on air conditioner operation data, spatial layout data and system topology data. The topology construction module is used to take each indoor air conditioner unit as a graph node based on the node feature data and the coupling relationship data, and construct corresponding multi-type coupling edges according to the refrigerant system coupling relationship, horizontal spatial coupling relationship and vertical spatial coupling relationship in the coupling relationship data to obtain the topology graph of the multi-split air conditioner group. The feature extraction module is used to dynamically update the edge weights of various types of coupling edges based on the topology map of the multi-split air conditioning group and the node feature data to obtain a dynamic adjacency matrix, and to perform graph attention spatial feature extraction and multi-scale temporal feature extraction based on the dynamic adjacency matrix and the node feature data to obtain spatial coupling feature data and temporal operation feature data. The gated fusion module is used to perform gated fusion processing on the spatial coupling feature data and the temporal running feature data to obtain spatiotemporal fusion feature data, and to perform hierarchical scheduling decision processing based on the spatiotemporal fusion feature data to obtain candidate collaborative scheduling data. The scheduling control module is used to perform constraint processing on the candidate coordinated scheduling data based on preset physical constraints to obtain a target coordinated scheduling strategy, and to perform scheduling control on the multi-split air conditioning group according to the target coordinated scheduling strategy.

[0011] In addition, to achieve the above objectives, this application also proposes a multi-split air conditioning group collaborative scheduling device, the device comprising: a memory, a processor, and a multi-split air conditioning group collaborative scheduling program stored in the memory and executable on the processor, the multi-split air conditioning group collaborative scheduling program being configured to implement the steps of the multi-split air conditioning group collaborative scheduling method as described above.

[0012] In addition, to achieve the above objectives, this application also proposes a storage medium storing a multi-split air conditioning group collaborative scheduling program, which, when executed by a processor, implements the steps of the multi-split air conditioning group collaborative scheduling method described above.

[0013] In addition, to achieve the above objectives, this application also proposes a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the multi-split air conditioning group collaborative scheduling method described above.

[0014] This application determines the node characteristic data corresponding to each indoor air conditioner unit and the coupling relationship data between the indoor air conditioners based on air conditioner operation data, spatial layout data, and system topology data. Based on the node characteristic data and coupling relationship data, each indoor air conditioner unit is treated as a graph node, and corresponding multi-type coupling edges are constructed according to the refrigerant system coupling relationship, horizontal spatial coupling relationship, and vertical spatial coupling relationship in the coupling relationship data to obtain a multi-split air conditioner group topology graph. Based on the multi-split air conditioner group topology graph and node characteristic data, the edge weights of each multi-type coupling edge are dynamically updated to obtain a dynamic adjacency matrix. Based on the dynamic adjacency matrix and node characteristic data, graph attention spatial feature extraction and multi-scale temporal feature extraction are performed to obtain spatial coupling feature data and temporal operation feature data. Gated fusion processing is performed on the spatial coupling feature data and temporal operation feature data to obtain spatiotemporal fusion feature data. Based on the spatiotemporal fusion feature data, hierarchical scheduling decision processing is performed to obtain candidate collaborative scheduling data. Based on preset physical constraints, the candidate collaborative scheduling data is constrained to obtain a target collaborative scheduling strategy, and the multi-split air conditioner group is scheduled and controlled according to the target collaborative scheduling strategy. This application establishes a unified data foundation for centralized building air conditioning control by determining node characteristic data and coupling relationship data based on air conditioning operation data, spatial layout data, and system topology data. Each indoor air conditioning unit is abstracted as a graph node, and multiple types of coupling edges are constructed by combining refrigerant system coupling relationships, horizontal spatial coupling relationships, and vertical spatial coupling relationships to form a topology graph of a multi-split air conditioning group, thereby representing the group association relationships between indoor air conditioning units. Furthermore, edge weights are dynamically updated based on the topology graph and node characteristic data to obtain a dynamic adjacency matrix, and spatial coupling characteristic data and temporal operation characteristic data are extracted, enabling the control process to simultaneously consider spatial associations and temporal changes. On this basis, candidate collaborative scheduling data is generated through gating fusion and hierarchical scheduling decisions, and constraint processing is performed using preset physical constraints to obtain an executable target collaborative scheduling strategy. This achieves collaborative scheduling control of multi-split air conditioning groups, improving the synergy of centralized building air conditioning control. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating the first embodiment of the multi-split air conditioning group collaborative scheduling method of this application; Figure 2 This is a schematic diagram of a sub-process in the second embodiment of the multi-split air conditioning group collaborative scheduling method of this application; Figure 3 This is a schematic diagram of a sub-process in the third embodiment of the multi-split air conditioning group collaborative scheduling method of this application; Figure 4 This is a schematic diagram of the module structure of the multi-split air conditioning group collaborative scheduling device according to an embodiment of this application; Figure 5This is a schematic diagram of the equipment structure of the hardware operating environment involved in the multi-split air conditioning group collaborative scheduling method in the embodiments of this application.

[0016] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0017] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of this application.

[0018] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0019] It should be noted that multi-split air conditioning systems are widely used in large buildings such as commercial buildings, hospitals, campuses, and hotels. They typically consist of one or more outdoor units and multiple indoor units connected to them. The indoor units are distributed across different rooms, floors, or functional areas and are linked to their corresponding outdoor units via refrigerant loops. In actual operation, multi-split air conditioning systems are affected not only by changes in the indoor and outdoor environment, occupant activity, and room load, but also by the refrigerant distribution within the same outdoor unit system and the heat transfer between adjacent spaces. Existing centralized building air conditioning control systems typically employ rule-based control methods, such as controlling on / off according to a fixed schedule, controlling the operation of a single air conditioner according to indoor temperature thresholds, or rotating operations according to preset groups. This type of control primarily focuses on independent control of individual air conditioners or fixed groups, failing to adequately consider the refrigerant system coupling relationships between different indoor units, the horizontal spatial heat coupling relationships between adjacent areas on the same floor, and the vertical spatial heat coupling relationships between floors. Therefore, when multiple areas experience simultaneous changes in cooling and heating loads, problems such as independent control strategies for different indoor air conditioning units, repeated cooling or heating in adjacent areas, and uncoordinated load distribution in outdoor unit systems can easily arise, leading to poor centralized control performance of building air conditioning systems. Thus, improving the synergy of centralized control of building air conditioning systems has become an urgent technical problem to be solved.

[0020] The main solution of this application is as follows: Based on air conditioning operation data, spatial layout data, and system topology data, determine the node characteristic data corresponding to each indoor air conditioning unit and the coupling relationship data between the indoor air conditioning units; based on the node characteristic data and coupling relationship data, treat each indoor air conditioning unit as a graph node, and construct corresponding multi-type coupling edges according to the refrigerant system coupling relationship, horizontal spatial coupling relationship, and vertical spatial coupling relationship in the coupling relationship data to obtain the topology graph of the multi-split air conditioning group; based on the topology graph of the multi-split air conditioning group and the node characteristic data, dynamically update the edge weights of each multi-type coupling edge to obtain a dynamic adjacency matrix, and extract graph attention spatial features and multi-scale temporal features based on the dynamic adjacency matrix and the node characteristic data to obtain spatial coupling feature data and temporal operation feature data; perform gating fusion processing on the spatial coupling feature data and temporal operation feature data to obtain spatiotemporal fusion feature data, and perform hierarchical scheduling decision processing based on the spatiotemporal fusion feature data to obtain candidate collaborative scheduling data; perform constraint processing on the candidate collaborative scheduling data based on preset physical constraints to obtain the target collaborative scheduling strategy, and perform scheduling control on the multi-split air conditioning group according to the target collaborative scheduling strategy.

[0021] This application establishes a unified data foundation for centralized building air conditioning control by determining node characteristic data and coupling relationship data based on air conditioning operation data, spatial layout data, and system topology data. Each indoor air conditioning unit is abstracted as a graph node, and multiple types of coupling edges are constructed by combining refrigerant system coupling relationships, horizontal spatial coupling relationships, and vertical spatial coupling relationships to form a topology graph of a multi-split air conditioning group, thereby representing the group association relationships between indoor air conditioning units. Furthermore, edge weights are dynamically updated based on the topology graph and node characteristic data to obtain a dynamic adjacency matrix, and spatial coupling characteristic data and temporal operation characteristic data are extracted, enabling the control process to simultaneously consider spatial associations and temporal changes. On this basis, candidate collaborative scheduling data is generated through gating fusion and hierarchical scheduling decisions, and constraint processing is performed using preset physical constraints to obtain an executable target collaborative scheduling strategy. This achieves collaborative scheduling control of multi-split air conditioning groups, improving the synergy of centralized building air conditioning control.

[0022] It should be noted that the executing entity of the method in this embodiment can be a computing service device with data processing, network communication, and program execution functions, or it can be the aforementioned multi-split air conditioning group collaborative scheduling device with the same or similar functions. This embodiment and the following embodiments will be described using a multi-split air conditioning group collaborative scheduling device as an example.

[0023] Based on this, a first embodiment of the multi-split air conditioning group collaborative scheduling method of this application is proposed. Please refer to [the relevant documentation]. Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the multi-split air conditioning group collaborative scheduling method of this application.

[0024] In this embodiment, the method includes the following steps: S1: Based on air conditioning operation data, spatial layout data, and system topology data, determine the node characteristic data corresponding to each air conditioning indoor unit and the coupling relationship data between air conditioning indoor units; S2: Based on the node feature data and the coupling relationship data, each indoor air conditioner unit is taken as a graph node, and corresponding multi-type coupling edges are constructed according to the refrigerant system coupling relationship, horizontal spatial coupling relationship and vertical spatial coupling relationship in the coupling relationship data to obtain the topology graph of the multi-split air conditioner group; It should be noted that air conditioning operation data refers to data reflecting the current or historical operating status of each indoor air conditioning unit, which may include information such as temperature status, operating mode, fan speed status, and energy consumption status; spatial layout data refers to data representing the floor location, room location, and adjacent relationships of each indoor air conditioning unit in the building; system topology data refers to data representing the connection relationships, ownership relationships, and refrigerant system relationships between the outdoor and indoor air conditioning units; node characteristic data refers to characteristic data describing the operating status and system attributes of the corresponding indoor air conditioning unit when each indoor air conditioning unit is treated as a node in a graph structure; coupling relationship data refers to data describing the relationships between different indoor air conditioning units, including refrigerant system coupling relationships. Horizontal and vertical spatial coupling relationships; graph nodes refer to node objects in the graph structure that correspond to indoor air conditioning units; multi-type coupling edges refer to edges used to connect different graph nodes and distinguish different coupling types; refrigerant system coupling relationship refers to the coupling relationship formed between indoor air conditioning units belonging to the same outdoor unit system due to refrigerant circuits or outdoor unit system associations; horizontal spatial coupling relationship refers to the spatial thermal association relationship formed between indoor air conditioning units in adjacent spaces on the same floor; vertical spatial coupling relationship refers to the spatial thermal association relationship formed between indoor air conditioning units in corresponding spaces on upper and lower floors; multi-split air conditioning group topology graph refers to the graph structure data composed of graph nodes corresponding to indoor air conditioning units and multi-type coupling edges between graph nodes.

[0025] Specifically, the air conditioning operation data is first analyzed to extract information such as temperature status, operating status, fan speed status, energy consumption status, and system status for each indoor unit. This information is then grouped according to the equipment identifier of the indoor unit, giving each indoor unit a characteristic description corresponding to its operating status. Next, the system topology data is used to determine the affiliation between each indoor unit and the outdoor unit system, and the refrigerant system associations between different indoor units within the same outdoor unit system are identified. Simultaneously, spatial layout data is used to determine the floor, room location, adjacent relationships on the same floor, and correspondence between floors above and below each indoor unit, thus obtaining the horizontal and vertical spatial associations between the indoor units. Through the above processing, the air conditioning operating status, spatial location relationships, and system connection relationships are uniformly organized, ultimately determining the node characteristic data corresponding to each indoor unit and the coupling relationship data between the indoor units.

[0026] After obtaining node feature data and coupling relationship data, each indoor air conditioner unit is mapped to a graph node, and the corresponding node feature data is associated with that graph node, enabling the graph node to represent the operating status and system attributes of the corresponding indoor air conditioner unit. Then, based on the refrigerant system coupling relationship in the coupling relationship data, refrigerant system coupling edges are constructed between graph nodes belonging to the same outdoor unit system; based on horizontal spatial coupling relationships, horizontal spatial coupling edges are constructed between graph nodes on the same floor and with spatial adjacency; and based on vertical spatial coupling relationships, vertical spatial coupling edges are constructed between graph nodes at corresponding positions on upper and lower floors. Finally, edge type labeling and connection relationship integration are performed on the above different types of coupling edges, so that each graph node and its corresponding multi-type coupling edges form a unified graph structure, thus obtaining the topology graph of the multi-split air conditioning group.

[0027] By determining node characteristic data and coupling relationship data based on air conditioning operation data, spatial layout data, and system topology data, the operating status, spatial location, and system affiliation of each air conditioning unit can be uniformly expressed. Furthermore, each air conditioning unit is treated as a graph node, and multiple types of coupling edges are constructed based on refrigerant system coupling relationships, horizontal spatial coupling relationships, and vertical spatial coupling relationships. This organizes the originally dispersed air conditioning units into a group topology structure with various relationships. Therefore, subsequent centralized control no longer relies solely on the independent state of a single air conditioner but can identify system and spatial relationships between different air conditioning units based on the multi-split air conditioning group topology graph. This provides a structured foundation for subsequent dynamic adjacency matrix construction, spatial feature extraction, and collaborative scheduling control, thereby improving the completeness of group state representation and the synergy of scheduling control during centralized building air conditioning control.

[0028] S3: Based on the topology of the multi-split air conditioning group and the node feature data, the edge weights of each type of coupling edge are dynamically updated to obtain a dynamic adjacency matrix. Based on the dynamic adjacency matrix and the node feature data, graph attention spatial feature extraction and multi-scale temporal feature extraction are performed to obtain spatial coupling feature data and temporal operation feature data. S4: Perform gated fusion processing on the spatial coupling feature data and the temporal operation feature data to obtain spatiotemporal fusion feature data, and perform hierarchical scheduling decision processing based on the spatiotemporal fusion feature data to obtain candidate collaborative scheduling data; S5: Based on preset physical constraints, the candidate coordinated scheduling data is constrained to obtain the target coordinated scheduling strategy, and the multi-split air conditioning group is scheduled and controlled according to the target coordinated scheduling strategy.

[0029] It should be noted that edge weights refer to data used to characterize the connection strength of multiple types of coupled edges; the dynamic adjacency matrix refers to a node connection relationship matrix generated based on the changes in edge weights of various types of coupled edges, used to represent the real-time changing correlation strength between different air conditioner indoor units; graph attention spatial feature extraction refers to the process of performing attention-weighted aggregation on the features of related air conditioner indoor unit nodes based on the dynamic adjacency matrix to obtain spatial correlation features; multi-scale temporal feature extraction refers to the process of analyzing the historical changes of node feature data from different time scales to obtain the characteristics of air conditioner operating status changes over time; spatial coupling feature data refers to feature data used to characterize the spatial correlation influence between air conditioner indoor units; temporal operation feature data refers to feature data used to characterize the changing pattern of air conditioner indoor unit operating status over time; gating fusion... The process of fusion processing refers to the adaptive determination of the fusion ratio between spatial coupling feature data and temporal operational feature data based on the degree of correlation between the two. Spatiotemporal fusion feature data refers to the comprehensive feature data formed by fusing spatial correlation information and temporal change information. Hierarchical scheduling decision processing refers to the process of generating scheduling results according to the hierarchy of outdoor unit system load allocation, indoor unit parameter adjustment, and time-series execution planning. Candidate collaborative scheduling data refers to the scheduling data obtained through hierarchical scheduling decision processing that still needs to be constrained. Preset physical constraints refer to the conditions used to limit the scheduling results to meet the actual operating boundaries of the air conditioning system. Target collaborative scheduling strategy refers to the scheduling strategy used to control the operation of multi-split air conditioning groups after constraint processing. Scheduling control refers to the process of controlling the operation of the outdoor unit system and indoor air conditioning units according to the target collaborative scheduling strategy.

[0030] Specifically, the coupling strength between different indoor air conditioning units is dynamically analyzed based on the connection objects, edge types, and corresponding node feature data of various coupling edges in the topology graph of the multi-split air conditioning group. Specifically, the edge weights of refrigerant system coupling edges, horizontal spatial coupling edges, and vertical spatial coupling edges are updated by combining the temperature state, operating state of each indoor air conditioning unit, and state differences between adjacent nodes, so that the edge weights can reflect the strength of the association between different nodes under the current operating state. Then, the updated edge weights of each type of coupling edge are organized into a dynamic adjacency matrix, and the range and influence relationship of neighboring nodes of each indoor air conditioning unit node are determined based on this dynamic adjacency matrix. On this basis, attention weights are calculated and weighted aggregation is performed on the node feature data of neighboring nodes through graph attention spatial feature extraction to obtain spatial coupling feature data that can characterize the spatial association influence between indoor air conditioning units. Simultaneously, historical node feature sequences of each indoor air conditioning unit are constructed based on the node feature data, and time-series feature extraction is performed on the historical node feature sequences at different time scales to obtain time-series operational feature data reflecting short-term changes, medium-term trends, and long-term change patterns in air conditioning operation.

[0031] After obtaining spatial coupling feature data and temporal operational feature data, dimensional matching, association coding, and gating fusion processing are performed on the two to adaptively determine the proportion of spatial and temporal information in the fusion process according to different operating states, resulting in spatiotemporal fusion feature data for each indoor air conditioning unit. Subsequently, hierarchical scheduling decision processing is performed based on the spatiotemporal fusion feature data. First, the group load is allocated at the outdoor unit system level. Then, the indoor unit parameter adjustment content is determined by combining the spatiotemporal fusion features of each indoor air conditioning unit, and a corresponding time-series execution plan is generated to obtain candidate collaborative scheduling data. Finally, the feasibility of the candidate collaborative scheduling data is verified and the constraints are corrected based on preset physical constraints, ensuring that the candidate collaborative scheduling data meets the operational boundary requirements such as outdoor unit capacity, temperature regulation, operating sequence, and system balance, thereby obtaining the target collaborative scheduling strategy. Based on the target collaborative scheduling strategy, corresponding scheduling control commands are issued to the multi-split air conditioning group to realize the scheduling control of the multi-split air conditioning group.

[0032] By dynamically updating the edge weights of various coupling edges based on the topology map and node feature data of multi-split air conditioning groups and generating a dynamic adjacency matrix, the correlation between indoor air conditioning units can be adjusted according to changes in operating status. Furthermore, spatial coupling feature data and temporal operating feature data are extracted based on the dynamic adjacency matrix and node feature data, ensuring that scheduling criteria simultaneously include the spatial correlation influence between indoor air conditioning units and the temporal variation patterns of operating status. On this basis, spatiotemporal fusion feature data is formed through gating fusion processing, and hierarchical scheduling decisions are made based on this spatiotemporal fusion feature data, enabling candidate collaborative scheduling data to reflect the hierarchical control relationship between the outdoor unit system and the indoor air conditioning units. Finally, preset physical constraints are used to constrain the candidate collaborative scheduling data to obtain an executable target collaborative scheduling strategy. This allows the centralized control process of multi-split air conditioning groups to take into account dynamic coupling representation, temporal operation analysis, hierarchical collaborative decision-making, and actual operating constraints, thereby improving the coordination, adaptability, and execution reliability of centralized building air conditioning control.

[0033] This embodiment determines the node feature data corresponding to each indoor air conditioner unit and the coupling relationship data between the indoor air conditioners based on air conditioner operation data, spatial layout data, and system topology data. Based on the node feature data and coupling relationship data, each indoor air conditioner unit is treated as a graph node, and corresponding multi-type coupling edges are constructed according to the refrigerant system coupling relationship, horizontal spatial coupling relationship, and vertical spatial coupling relationship in the coupling relationship data to obtain the topology graph of the multi-split air conditioner group. Based on the topology graph of the multi-split air conditioner group and the node feature data, the edge weights of each multi-type coupling edge are dynamically updated to obtain a dynamic adjacency matrix. Based on the dynamic adjacency matrix and the node feature data, graph attention spatial feature extraction and multi-scale temporal feature extraction are performed to obtain spatial coupling feature data and temporal operation feature data. Gated fusion processing is performed on the spatial coupling feature data and temporal operation feature data to obtain spatiotemporal fusion feature data. Hierarchical scheduling decision processing is performed based on the spatiotemporal fusion feature data to obtain candidate collaborative scheduling data. Based on preset physical constraints, the candidate collaborative scheduling data is constrained to obtain a target collaborative scheduling strategy, and the multi-split air conditioner group is scheduled and controlled according to the target collaborative scheduling strategy. This embodiment determines node characteristic data and coupling relationship data based on air conditioning operation data, spatial layout data, and system topology data, providing a unified data foundation for centralized control of building air conditioning. Each indoor air conditioning unit is abstracted as a graph node, and multiple types of coupling edges are constructed by combining refrigerant system coupling relationships, horizontal spatial coupling relationships, and vertical spatial coupling relationships to form a topology graph of the multi-split air conditioning group, thereby representing the group association relationships between the indoor air conditioning units. Furthermore, edge weights are dynamically updated based on the topology graph and node characteristic data to obtain a dynamic adjacency matrix, and spatial coupling characteristic data and temporal operation characteristic data are extracted, enabling the control process to simultaneously consider spatial associations and temporal changes. On this basis, candidate collaborative scheduling data is generated through gating fusion and hierarchical scheduling decisions, and constrained using preset physical constraints to obtain an executable target collaborative scheduling strategy. This achieves collaborative scheduling control of the multi-split air conditioning group, improving the synergy of centralized control of building air conditioning.

[0034] Based on the first embodiment described above, a second embodiment of the multi-split air conditioning group collaborative scheduling method of this application is proposed. Please refer to... Figure 2 , Figure 2 This is a schematic diagram of a sub-process in the second embodiment of the multi-split air conditioning group collaborative scheduling method of this application.

[0035] like Figure 2 As shown, in this embodiment, step S1 includes: S11: Analyze the operating parameters of the air conditioner operating data to determine the temperature status data, operating status data and energy consumption status data corresponding to each air conditioner indoor unit, and generate the operating characteristic data corresponding to each air conditioner indoor unit based on the temperature status data, the operating status data and the energy consumption status data. S12: Perform outdoor unit attribution analysis on the system topology data to determine the outdoor unit system to which each indoor air conditioner unit belongs and the refrigerant association information of each indoor air conditioner unit in the corresponding outdoor unit system, and determine the refrigerant system coupling relationship between indoor air conditioners based on the outdoor unit system and the refrigerant association information. S13: Based on the spatial layout data, determine the floor position relationship and spatial adjacency relationship of each air conditioner indoor unit, and determine the horizontal spatial coupling relationship and vertical spatial coupling relationship between the air conditioner indoor units according to the floor position relationship and the spatial adjacency relationship. Also, associate the operating characteristic data with the corresponding air conditioner indoor unit to obtain the node characteristic data corresponding to each air conditioner indoor unit, and obtain the coupling relationship data between the air conditioner indoor units according to the refrigerant system coupling relationship, the horizontal spatial coupling relationship and the vertical spatial coupling relationship.

[0036] It should be noted that operational parameter parsing refers to classifying and extracting fields from the collected air conditioning operational data to obtain operational parameters that can be used for subsequent modeling; temperature status data refers to data reflecting the temperature-related status of the air conditioning indoor unit, including set temperature, actual return air temperature, and supply air temperature; operational status data refers to data reflecting the operating mode of the air conditioning indoor unit, including operating mode, fan speed level, and start / stop status; energy consumption status data refers to data reflecting the energy consumption of the air conditioning indoor unit, including real-time power and operating current; and operational characteristic data refers to data formed by combining temperature status data, operational status data, and energy consumption status data. The system comprises characteristic data describing the operating status of a single indoor air conditioner unit; outdoor unit attribution resolution refers to the process of determining which outdoor unit system each indoor air conditioner unit is connected to or belongs to based on system topology data; an outdoor unit system refers to a system unit composed of the same outdoor air conditioner unit and multiple indoor air conditioners connected to it; refrigerant association information refers to information related to the sharing of refrigerant circuits, refrigerant distribution, or cooling and heating capacity of the indoor air conditioner unit in the corresponding outdoor unit system; floor location relationship refers to the distribution or correspondence of each indoor air conditioner unit on different floors in a building; spatial adjacency relationship refers to the physical adjacency relationship between the areas where different indoor air conditioner units are located on the same floor.

[0037] First, the air conditioning operation data is analyzed to extract temperature status data, operating status data, and energy consumption status data corresponding to each indoor unit, based on the unit's equipment identifier. Temperature status data characterizes the temperature control status of the area where the indoor unit is located; operating status data characterizes the current operating mode and fan speed of the indoor unit; and energy consumption status data characterizes the power or current consumption during operation. Subsequently, these different types of operating parameters are normalized, organized, and correlated to form a unified operating status description for the same indoor unit, thereby generating corresponding operating characteristic data for each indoor unit.

[0038] Then, the system topology data is analyzed to determine the outdoor unit affiliation, identifying the outdoor unit system to which each indoor unit belongs. Furthermore, the refrigerant association information of each indoor unit within its corresponding outdoor unit system is determined to identify the refrigerant system coupling relationship between different indoor units within the same outdoor unit system, formed by sharing the outdoor unit and refrigerant circuit. Based on this, the floor, area, and adjacent area relationships of each indoor unit are determined using spatial layout data. Vertical spatial coupling relationships between corresponding areas on upper and lower floors are identified based on floor location, and horizontal spatial coupling relationships between adjacent areas on the same floor are identified based on spatial adjacency. Finally, the operating characteristic data of each indoor unit is associated with its corresponding indoor unit to obtain the node characteristic data for each indoor unit. The refrigerant system coupling relationship, horizontal spatial coupling relationship, and vertical spatial coupling relationship are then integrated to obtain the coupling relationship data between the indoor units.

[0039] By analyzing the operating parameters of the air conditioning operation data, the temperature, operating status, and energy consumption status of each indoor unit can be organized into unified operating characteristic data, providing an accurate basis for subsequent processing based on individual unit status. By analyzing the outdoor unit affiliation relationships in the system topology data, the refrigerant system coupling relationships between indoor units within the same outdoor unit system can be identified. Furthermore, by determining floor location relationships and spatial adjacency relationships based on spatial layout data, the horizontal and vertical spatial coupling relationships between indoor units can be further identified. Thus, this step unifies the operating status of indoor units, outdoor unit affiliation relationships, and spatial association relationships into node characteristic data and coupling relationship data, providing a data foundation for subsequently constructing a multi-split air conditioning group topology map, thereby improving the completeness of group relationship representation and the accuracy of collaborative scheduling in centralized building air conditioning control.

[0040] Based on the first embodiment described above, in this embodiment, step S2 includes: S21: Based on the node feature data, perform node mapping processing on each indoor air conditioner unit to determine the graph nodes that correspond one-to-one with each indoor air conditioner unit, and associate each graph node with the corresponding node feature data to obtain a set of indoor air conditioner unit nodes. S22: Based on the refrigerant system coupling relationship in the coupling relationship data, perform refrigerant association matching on the graph nodes belonging to the same outdoor unit system in the set of indoor air conditioning units to construct refrigerant system coupling edges, and based on the horizontal spatial coupling relationship and the vertical spatial coupling relationship, construct horizontal spatial coupling edges and vertical spatial coupling edges respectively on the graph nodes with spatial thermal coupling relationship in the set of indoor air conditioning units. S23: Perform edge type marking and node connection relationship integration on the coupling edges of the refrigerant system, the horizontal space coupling edges, and the vertical space coupling edges to obtain a multi-type coupling edge set corresponding to the set of indoor air conditioning unit nodes, and generate the topology map of the multi-split air conditioning group based on the set of indoor air conditioning unit nodes and the multi-type coupling edge set.

[0041] It should be noted that node mapping processing refers to the process of converting actual indoor air conditioning units into node objects in a graph structure, so that each indoor air conditioning unit has a corresponding graph representation; the indoor air conditioning unit node set refers to the set of graph nodes formed by mapping each indoor air conditioning unit; refrigerant association matching refers to the process of determining whether different graph nodes are associated with the same refrigerant system based on the outdoor unit system to which the indoor air conditioning unit belongs and the refrigerant connection relationship; refrigerant system coupling edge refers to the edge used to connect graph nodes belonging to the same outdoor unit system; horizontal spatial coupling edge refers to the edge used to connect graph nodes with spatial thermal coupling relationship on the same floor; vertical spatial coupling edge refers to the edge used to connect graph nodes with spatial thermal coupling relationship between upper and lower floors; edge type labeling refers to the process of assigning corresponding type labels to different coupling edges; node connection relationship integration refers to the process of unifying and organizing the node connection relationships corresponding to various coupling edges; multi-type coupling edge set refers to the edge set composed of refrigerant system coupling edges, horizontal spatial coupling edges, and vertical spatial coupling edges.

[0042] Specifically, firstly, node mapping processing is performed on each indoor air conditioner unit based on node feature data. That is, according to the equipment identifier, region, or system number of the indoor air conditioner unit, each indoor air conditioner unit is converted into a graph node in the graph structure, maintaining a one-to-one correspondence between indoor air conditioner units and graph nodes. Subsequently, the node feature data corresponding to each indoor air conditioner unit is associated with the corresponding graph node, so that each graph node not only represents an indoor air conditioner unit, but also carries feature information such as the operating status, temperature status, energy consumption status, and system attributes of that indoor air conditioner unit, thereby forming a set of indoor air conditioner unit nodes.

[0043] After obtaining the set of indoor unit nodes, the system further performs refrigerant association matching on graph nodes belonging to the same outdoor unit system based on the refrigerant system coupling relationship in the coupling relationship data, and constructs refrigerant system coupling edges between successfully matched graph nodes. Simultaneously, based on horizontal spatial coupling relationships, horizontal spatial coupling edges are constructed between graph nodes on the same floor that have spatial thermal associations, and based on vertical spatial coupling relationships, vertical spatial coupling edges are constructed between graph nodes in corresponding areas of upper and lower floors that have spatial thermal associations. Finally, the refrigerant system coupling edges, horizontal spatial coupling edges, and vertical spatial coupling edges are labeled with edge types, and the graph node relationships connected by each type of edge are uniformly integrated to obtain a multi-type coupling edge set corresponding to the set of indoor unit nodes. Thus, a multi-split air conditioning group topology graph is generated based on the set of indoor unit nodes and the multi-type coupling edge set.

[0044] By mapping individual indoor air conditioning units based on their feature data, the dispersed units can be uniformly converted into node objects in a graph structure, with each node carrying the feature information of its corresponding indoor unit. Furthermore, different types of coupling edges are constructed based on refrigerant system coupling relationships, horizontal spatial coupling relationships, and vertical spatial coupling relationships. This allows refrigerant associations within the same outdoor unit system, thermal associations between spaces on the same floor, and thermal associations between floors to be expressed in the topology. Finally, by integrating edge type labels and node connection relationships, a multi-split air conditioning group topology graph containing multiple types of coupling edges is formed. Thus, this step transforms the multi-split air conditioning group from a simple set of devices into a graph structure with node features and multiple types of connection relationships. This provides a structured foundation for subsequent dynamic edge weight updates, spatial feature extraction, and collaborative scheduling decisions, thereby improving the ability of centralized building air conditioning control to represent the relationships within the air conditioning group.

[0045] This embodiment determines the node feature data corresponding to each indoor air conditioner unit and the coupling relationship data between the indoor air conditioners based on air conditioner operation data, spatial layout data, and system topology data. Based on the node feature data and coupling relationship data, each indoor air conditioner unit is treated as a graph node, and corresponding multi-type coupling edges are constructed according to the refrigerant system coupling relationship, horizontal spatial coupling relationship, and vertical spatial coupling relationship in the coupling relationship data to obtain the topology graph of the multi-split air conditioner group. Based on the topology graph of the multi-split air conditioner group and the node feature data, the edge weights of each multi-type coupling edge are dynamically updated to obtain a dynamic adjacency matrix. Based on the dynamic adjacency matrix and the node feature data, graph attention spatial feature extraction and multi-scale temporal feature extraction are performed to obtain spatial coupling feature data and temporal operation feature data. Gated fusion processing is performed on the spatial coupling feature data and temporal operation feature data to obtain spatiotemporal fusion feature data. Hierarchical scheduling decision processing is performed based on the spatiotemporal fusion feature data to obtain candidate collaborative scheduling data. Based on preset physical constraints, the candidate collaborative scheduling data is constrained to obtain a target collaborative scheduling strategy, and the multi-split air conditioner group is scheduled and controlled according to the target collaborative scheduling strategy. This embodiment determines node characteristic data and coupling relationship data based on air conditioning operation data, spatial layout data, and system topology data, providing a unified data foundation for centralized control of building air conditioning. Each indoor air conditioning unit is abstracted as a graph node, and multiple types of coupling edges are constructed by combining refrigerant system coupling relationships, horizontal spatial coupling relationships, and vertical spatial coupling relationships to form a topology graph of the multi-split air conditioning group, thereby representing the group association relationships between the indoor air conditioning units. Furthermore, edge weights are dynamically updated based on the topology graph and node characteristic data to obtain a dynamic adjacency matrix, and spatial coupling characteristic data and temporal operation characteristic data are extracted, enabling the control process to simultaneously consider spatial associations and temporal changes. On this basis, candidate collaborative scheduling data is generated through gating fusion and hierarchical scheduling decisions, and constrained using preset physical constraints to obtain an executable target collaborative scheduling strategy. This achieves collaborative scheduling control of the multi-split air conditioning group, improving the synergy of centralized control of building air conditioning.

[0046] Based on the second embodiment described above, a third embodiment of the multi-split air conditioning group collaborative scheduling method of this application is proposed. Please refer to... Figure 3 , Figure 3 This is a schematic diagram of a sub-process in the third embodiment of the multi-split air conditioning group collaborative scheduling method of this application.

[0047] In this embodiment, step S3 includes: S31: Based on the edge type information, node connection relationship and temperature status information in the node feature data of each multi-type coupling edge in the topology diagram of the multi-split air conditioning group, the coupling strength corresponding to each multi-type coupling edge is dynamically characterized to obtain the updated edge weight corresponding to each multi-type coupling edge, and the dynamic adjacency matrix is ​​generated according to the updated edge weight. S32: Based on the dynamic adjacency matrix, determine the set of neighboring nodes corresponding to each graph node, and according to the node feature data, the edge type information and the updated edge weight, perform graph attention calculation and feature aggregation processing on the spatial influence weight between each graph node and its corresponding neighboring node to obtain the spatial coupling feature data corresponding to each air conditioner indoor unit. S33: Based on the node feature data, construct the historical node feature sequence corresponding to each air conditioner indoor unit, and perform multi-scale causal convolution processing and scale fusion processing on the historical node feature sequence to obtain the time operation feature data corresponding to each air conditioner indoor unit.

[0048] It should be noted that edge type information refers to information used to distinguish the categories of different types of coupling edges, such as refrigerant system coupling edges, horizontal space coupling edges, or vertical space coupling edges; node connection relationship refers to the correspondence between the two graph nodes connected by each coupling edge; temperature state information refers to the data in the node feature data used to reflect the temperature state of the air conditioner indoor unit, such as the set temperature, return air temperature, and supply air temperature; coupling strength refers to the degree of mutual influence between two air conditioner indoor units due to system connection or spatial heat transfer; dynamic characterization processing refers to the process of updating the coupling strength in real time or periodically based on the current node characteristics and edge relationship changes; updated edge weight refers to the weight data obtained after dynamic characterization processing, used to represent the current coupling edge connection strength; and the dynamic adjacency matrix refers to the matrix generated based on the updated edge weights of each coupling edge, used to describe... A matrix representing the current connection relationships and connection strengths between graph nodes; a neighbor node set refers to the set of other graph nodes that are connected to a given graph node by a coupling edge; spatial influence weight refers to the weight corresponding to the spatial influence of a neighbor node on the current graph node; graph attention calculation refers to the calculation process that adaptively determines the degree of influence of neighbor nodes based on node features, edge types, and edge weights; feature aggregation processing refers to the process of aggregating the features of neighbor nodes into the current graph node according to their spatial influence weights; historical node feature sequence refers to the set of node feature data arranged in chronological order; multi-scale causal convolution processing refers to the convolution processing that extracts temporal features from the historical node feature sequence at different time scales, using only the data from the current and past times; scale fusion processing refers to the process of merging the temporal features extracted from different time scales.

[0049] Specifically, the process begins by reading the edge type information and node connection relationships corresponding to each type of coupling edge in the topology diagram of the multi-split air conditioning group. Combined with the temperature state information in the node feature data, the actual coupling state represented by each type of coupling edge is analyzed. For refrigerant system coupling edges, the coupling basis between nodes in the same outdoor unit system can be determined based on the operating status of its associated outdoor unit system and related nodes. For horizontal and vertical spatial coupling edges, the spatial thermal correlation changes can be determined by considering the temperature state differences between adjacent nodes. Therefore, the coupling strength corresponding to each type of coupling edge is dynamically characterized to obtain the updated edge weights for each type of coupling edge at the current control moment. These updated edge weights are then written into the corresponding matrix positions according to the connection relationships between the nodes, generating a dynamic adjacency matrix.

[0050] After obtaining the dynamic adjacency matrix, the set of neighboring nodes corresponding to each graph node is determined based on the dynamic adjacency matrix. Then, for the current graph node and its neighboring nodes, graph attention is calculated by combining node feature data, edge type information, and updated edge weights to determine the spatial influence weight of different neighboring nodes on the current graph node. Subsequently, feature aggregation processing is performed on the features of the neighboring nodes according to the spatial influence weights to obtain the spatial coupling feature data corresponding to each air conditioner indoor unit. Simultaneously, based on the node feature data, a historical node feature sequence corresponding to each air conditioner indoor unit is constructed in chronological order. Multi-scale causal convolution processing is then performed on the historical node feature sequence to extract operational change features at different time ranges. Finally, scale fusion processing is performed on the temporal features at each scale to obtain the temporal operational feature data corresponding to each air conditioner indoor unit.

[0051] By dynamically representing the coupling strength of various coupling edges based on edge type information, node connection relationships, and temperature status information in the node feature data of the multi-split air conditioning group topology graph, the connection strength between graph nodes can be updated according to the current operating status and temperature status of the indoor air conditioning units, thus obtaining a dynamic adjacency matrix that reflects the current coupling state of the group. Further graph attention calculation and feature aggregation processing based on the dynamic adjacency matrix can extract spatial coupling feature data matching the degree of spatial association from the neighboring nodes of each graph node. Simultaneously, by constructing historical node feature sequences and performing multi-scale causal convolution and scale fusion processing, the changing features of the indoor air conditioning unit operating status can be extracted from different time scales, obtaining temporal operating feature data. Therefore, this step simultaneously forms a feature foundation reflecting the spatial coupling relationship and temporal operating changes of the air conditioning group, providing a more complete spatiotemporal representation basis for subsequent gating fusion and hierarchical scheduling decisions, thereby improving the building air conditioning centralized control's ability to identify dynamic group states and the accuracy of collaborative scheduling.

[0052] Based on the second embodiment described above, in this embodiment, step S4 includes: S41: Perform feature dimension matching and association encoding processing on the spatial coupling feature data and the time operation feature data to obtain spatiotemporal candidate representation data, and determine the gate control weight data corresponding to each air conditioner indoor unit based on the spatiotemporal candidate representation data; S42: Based on the gating weight data, the spatial coupling feature data and the time operation feature data are weighted and fused to obtain the spatiotemporal fusion feature data corresponding to each air conditioner indoor unit. According to the affiliation relationship between the air conditioner indoor unit and the outdoor unit system, the spatiotemporal fusion feature data is aggregated at the outdoor unit system level to obtain the outdoor unit system aggregated feature data. S43: Based on the aggregated feature data of the external system, a global load allocation decision is made to obtain the load allocation data of the external system. Based on the load allocation data of the external system and the spatiotemporal fusion feature data, an internal parameter adjustment decision and a timing execution planning process are performed to obtain candidate collaborative scheduling data.

[0053] It should be noted that feature dimension matching refers to the process of unifying the dimensions and aligning the formats of feature data from different sources; association encoding refers to the process of encoding and expressing the correspondence between spatial features and temporal features; spatiotemporal candidate representation data refers to candidate fusion features formed by matching and encoding spatially coupled feature data and temporally dynamic feature data; gating weight data refers to weight data used to control the proportion of spatially coupled feature data and temporally dynamic feature data in the fusion process; weighted fusion processing refers to the process of combining different features according to the gating weight data; spatiotemporal fusion feature data refers to feature data formed after fusing spatial correlation information and temporal change information; and outdoor unit system-level aggregation processing refers to the aggregation of features within the same outdoor unit system according to the outdoor unit system to which the indoor unit belongs. The process of summarizing and expressing the integrated feature data of the air conditioning system; the aggregated feature data of the outdoor unit system refers to the feature data used to characterize the overall operating status and load association status of the corresponding outdoor unit system; the global load allocation decision refers to the process of determining the load allocation of each outdoor unit system at the outdoor unit system level; the load allocation data of the outdoor unit system refers to the data used to represent the target load allocation result of each outdoor unit system; the indoor unit parameter adjustment decision refers to the process of determining the adjustment amount for the set temperature, fan speed and other operating parameters of each air conditioning indoor unit; the time-series execution planning process refers to the process of organizing the adjustment decision into a step-by-step execution plan in multiple subsequent control periods; the candidate collaborative scheduling data refers to the candidate scheduling data generated before the constraint processing, used to characterize the load allocation of the outdoor unit system, the indoor unit parameter adjustment and the time-series execution plan.

[0054] Specifically, firstly, feature dimension matching is performed on spatial coupling feature data and temporal operation feature data to ensure consistency in data dimensions, node correspondence, and expression format between the two types of features, avoiding incompatibility during subsequent fusion due to different feature sources. Then, the matched spatial coupling feature data and temporal operation feature data undergo associative encoding processing to express the spatial association state and temporal operation changes of the same air conditioner indoor unit, obtaining spatiotemporal candidate representation data. Finally, based on the spatiotemporal candidate representation data, the gating weight data corresponding to each air conditioner indoor unit is determined, enabling each air conditioner indoor unit to form a weight basis for controlling the fusion ratio of spatial and temporal information under different operating states.

[0055] After obtaining the gating weight data, the spatial coupling feature data and temporal operation feature data are weighted and fused according to the gating weight data to obtain the spatiotemporal fusion feature data corresponding to each air conditioner indoor unit. This ensures that the feature expression of each air conditioner indoor unit simultaneously includes its spatial coupling relationship and temporal operation changes. Subsequently, based on the affiliation relationship between the air conditioner indoor unit and the outdoor unit system, the spatiotemporal fusion feature data belonging to the same outdoor unit system are aggregated at the outdoor unit system level to obtain the outdoor unit system aggregated feature data. Based on the outdoor unit system aggregated feature data, a global load allocation decision is made to determine the outdoor unit system load allocation data corresponding to each outdoor unit system. Furthermore, by combining the outdoor unit system load allocation data and the spatiotemporal fusion feature data corresponding to each air conditioner indoor unit, indoor unit parameter adjustment decisions and timing execution planning are performed respectively, ultimately obtaining candidate collaborative scheduling data.

[0056] By performing feature dimension matching and correlation encoding on spatial coupling feature data and temporal operation feature data, spatial and temporal information from different sources can be integrated into a unified representation. Further, based on the spatiotemporal candidate representation data, gating weight data is determined, and weighted fusion processing is performed accordingly. This allows the spatiotemporal fusion feature data corresponding to each indoor air conditioning unit to simultaneously reflect both spatial correlation status and temporal operation changes. Then, based on the attribution relationship between the indoor and outdoor air conditioning systems, outdoor system-level aggregation is performed. Global load allocation decisions, indoor unit parameter adjustment decisions, and sequential execution planning are then performed based on the aggregated outdoor system feature data. This ensures that the scheduling results simultaneously cover load allocation at the outdoor system level, parameter adjustment at the indoor unit level, and step-by-step execution arrangements at the temporal level. Thus, this step transforms the previously extracted spatiotemporal features into candidate collaborative scheduling data with hierarchical relationships and execution order, providing a decision-making basis for subsequent physical constraint processing and target collaborative scheduling strategy generation, thereby improving the collaborative scheduling capability in the centralized control of multi-split air conditioning groups.

[0057] In this embodiment, step S5 includes: S51: Parse the scheduling content of the candidate collaborative scheduling data to determine the load distribution data of the external system, the parameter adjustment data of the internal system, and the timing execution plan data, and establish constraint verification objects for the load distribution data of the external system, the parameter adjustment data of the internal system, and the timing execution plan data based on the preset physical constraints. S52: Based on the constraint verification object, perform feasibility verification on the candidate collaborative scheduling data for outdoor unit capacity constraints, temperature change rate constraints, minimum running time constraints, and refrigerant balance constraints, and perform constraint correction processing on the candidate collaborative scheduling data that does not meet the preset physical constraint conditions to obtain constrained collaborative scheduling data. S53: Generate a target collaborative scheduling strategy based on the constrained collaborative scheduling data, and issue load distribution instructions, parameter adjustment instructions, and timing execution instructions to the corresponding outdoor unit system and indoor air conditioning unit according to the target collaborative scheduling strategy, so as to perform scheduling control on the multi-split air conditioning group.

[0058] It should be noted that scheduling content parsing refers to the process of classifying and extracting different scheduling contents contained in candidate collaborative scheduling data; preset physical constraints refer to the hard limitations that multi-split air conditioning groups need to meet in actual operation; constraint verification objects refer to data objects established based on different scheduling contents for physical constraint judgment; outdoor unit capacity constraints refer to the limitations formed by the actual carrying capacity of the outdoor unit system on the load distribution results; temperature change rate constraints refer to the limitations that the set temperature adjustment range or change rate of the indoor unit of the air conditioner needs to meet; minimum operating time constraints refer to the minimum time limit that the indoor unit of the air conditioner or related equipment needs to meet during start-up, shutdown or operation switching; refrigerant balance constraints refer to... The cooling or heating demand of each indoor air conditioner unit under the same outdoor unit system needs to match the output capacity of the outdoor unit system; feasibility verification refers to the process of determining whether candidate collaborative scheduling data meets the preset physical constraints; constraint correction processing refers to the process of adjusting scheduling data that does not meet the preset physical constraints; constrained collaborative scheduling data refers to the collaborative scheduling data obtained after feasibility verification and constraint correction; target collaborative scheduling strategy refers to the final scheduling strategy used to actually control the operation of multi-split air conditioning groups; load distribution instructions, parameter adjustment instructions, and timing execution instructions refer to the load control, operating parameter control, and time-sharing execution control instructions issued to the outdoor unit system or indoor air conditioner units, respectively.

[0059] Specifically, the candidate coordinated scheduling data is first parsed, breaking down the content involving different control levels and execution objects to determine the outdoor unit system load allocation data, indoor unit parameter adjustment data, and time-series execution plan data. The outdoor unit system load allocation data represents the load arrangement at the outdoor unit system level, the indoor unit parameter adjustment data represents the adjustment of settings such as set temperature and fan speed at the indoor unit level, and the time-series execution plan data represents the step-by-step execution arrangement within subsequent control periods. Then, based on preset physical constraints, corresponding constraint verification objects are established for the outdoor unit system load allocation data, indoor unit parameter adjustment data, and time-series execution plan data, ensuring that different types of scheduling content can enter the corresponding constraint verification process.

[0060] After establishing the constraint verification objects, the feasibility of candidate collaborative scheduling data is verified based on these objects. Specifically, this includes verifying outdoor unit capacity constraints on outdoor unit system load allocation data, temperature change rate constraints and refrigerant balance constraints on indoor unit parameter adjustment data, and minimum running time constraints on timing execution plan data. When any candidate collaborative scheduling data does not meet the preset physical constraints, the corresponding load allocation results, parameter adjustment results, or execution plans are adjusted to meet the constraints, resulting in constrained collaborative scheduling data. Finally, a target collaborative scheduling strategy is generated based on the constrained collaborative scheduling data. According to the target collaborative scheduling strategy, load allocation instructions are issued to the corresponding outdoor unit systems, and parameter adjustment instructions and timing execution instructions are issued to the corresponding indoor air conditioning units, thereby achieving scheduling control of multi-split air conditioning groups.

[0061] By parsing the candidate coordinated scheduling data, the load allocation of the outdoor unit system, the adjustment of indoor unit parameters, and the timing execution plan can be processed as verifiable scheduling content. Furthermore, based on preset physical constraints, constraint verification objects are established, and the feasibility of outdoor unit capacity constraints, temperature change rate constraints, minimum operating time constraints, and refrigerant balance constraints are verified on the candidate coordinated scheduling data. This allows for the identification of scheduling content that does not meet physical operating limitations before the actual execution of the scheduling strategy. Constraint correction processing then yields constrained coordinated scheduling data, from which the target coordinated scheduling strategy and corresponding control commands are generated. Thus, this step transforms the previously generated candidate scheduling results into an executable control strategy that conforms to the actual operating conditions of multi-split air conditioning groups, providing a reliable constraint execution basis for centralized building air conditioning control, thereby improving the executability of scheduling control and the coordination of system operation.

[0062] This embodiment determines the node feature data corresponding to each indoor air conditioner unit and the coupling relationship data between the indoor air conditioners based on air conditioner operation data, spatial layout data, and system topology data. Based on the node feature data and coupling relationship data, each indoor air conditioner unit is treated as a graph node, and corresponding multi-type coupling edges are constructed according to the refrigerant system coupling relationship, horizontal spatial coupling relationship, and vertical spatial coupling relationship in the coupling relationship data to obtain the topology graph of the multi-split air conditioner group. Based on the topology graph of the multi-split air conditioner group and the node feature data, the edge weights of each multi-type coupling edge are dynamically updated to obtain a dynamic adjacency matrix. Based on the dynamic adjacency matrix and the node feature data, graph attention spatial feature extraction and multi-scale temporal feature extraction are performed to obtain spatial coupling feature data and temporal operation feature data. Gated fusion processing is performed on the spatial coupling feature data and temporal operation feature data to obtain spatiotemporal fusion feature data. Hierarchical scheduling decision processing is performed based on the spatiotemporal fusion feature data to obtain candidate collaborative scheduling data. Based on preset physical constraints, the candidate collaborative scheduling data is constrained to obtain a target collaborative scheduling strategy, and the multi-split air conditioner group is scheduled and controlled according to the target collaborative scheduling strategy. This embodiment determines node characteristic data and coupling relationship data based on air conditioning operation data, spatial layout data, and system topology data, providing a unified data foundation for centralized control of building air conditioning. Each indoor air conditioning unit is abstracted as a graph node, and multiple types of coupling edges are constructed by combining refrigerant system coupling relationships, horizontal spatial coupling relationships, and vertical spatial coupling relationships to form a topology graph of the multi-split air conditioning group, thereby representing the group association relationships between the indoor air conditioning units. Furthermore, edge weights are dynamically updated based on the topology graph and node characteristic data to obtain a dynamic adjacency matrix, and spatial coupling characteristic data and temporal operation characteristic data are extracted, enabling the control process to simultaneously consider spatial associations and temporal changes. On this basis, candidate collaborative scheduling data is generated through gating fusion and hierarchical scheduling decisions, and constrained using preset physical constraints to obtain an executable target collaborative scheduling strategy. This achieves collaborative scheduling control of the multi-split air conditioning group, improving the synergy of centralized control of building air conditioning.

[0063] In one embodiment, the specific steps of constructing a multi-split air conditioning topology model using the multi-split air conditioning group collaborative scheduling method of this application include: 1. Node definition and feature construction Each indoor air conditioning unit in the building is treated as a node in the graph G=(V, E). The feature vector of each node Each time step t includes the following parameters (collected via the air conditioning smart gateway): Temperature characteristics (temperature status data): Current set temperature Actual return air temperature air supply temperature .

[0064] Operating characteristics (operating characteristic data): Operating mode (Cooling=0 / Heating=1 / Air Supply=2), Fan Speed ​​Level (Low=1 / Medium=2 / High=3 / Auto=4).

[0065] Energy consumption characteristics (energy status data): real-time power Operating current .

[0066] System characteristics (system topology data): Number of the associated outdoor unit Refrigerant distribution ratio in the outdoor unit system .

[0067] 2. Construction of multiple edge types Construct the edge set E of the graph based on the three physical coupling relationships: Type 1 – Refrigerant System Coupling Edge: A fully connected edge is established between all indoor units within the same outdoor unit system. Edge Weight This indicates the strongest hardware-level coupling (shared refrigerant circuit and outdoor compressor capability).

[0068] Type 2 – Horizontal Spatial Coupling Edge: An edge is established between physically adjacent indoor units on the same floor. Edge weight. ,in The distance between the centers of the rooms where the two indoor units are located (in meters) is given. α and σ are learnable parameters (initial values ​​α=0.8, σ=10.0). This formula indicates that the thermal coupling strength decreases exponentially with increasing distance.

[0069] Type 3 – Vertical Spatial Coupling Edge: An edge is established between corresponding indoor units on adjacent floors. Edge weight. ,in The floor spacing is typically 3 to 4 meters, and β and γ are learnable parameters (initial values ​​β=0.5, γ=5.0).

[0070] 3. Construction of dynamic adjacency matrix The three types of edge weights mentioned above are merged into a unified adjacency matrix A(t), whose elements change dynamically over time:

[0071] in The weights are the three types of edges mentioned above (the maximum value is taken if there are multiple types of edges between nodes i and j), and ψ is the temperature difference attention function:

[0072] The meaning of this function is: when the absolute value of the temperature difference between adjacent nodes... When it is large (exceeds the threshold) The default temperature is 2℃. Strong heat conduction drives the graph neural network, increasing the adjacent weights and causing it to focus more on neighboring nodes with large temperature differences. The parameter k controls the steepness of the sigmoid function (default k=2.0).

[0073] In one embodiment, the multi-objective collaborative scheduling decision in the multi-split air conditioning group collaborative scheduling method of this application specifically includes the following steps: 1. Multi-objective optimization function Define three optimization objectives: Objective 1 – Minimize total energy consumption: ,in Let Δt be the power of the i-th indoor unit at time t, and Δt be the control step size (default 15 minutes).

[0074] Objective 2 – Minimize comfort deviation: Where ε is the acceptable temperature deviation tolerance (default ε = 1.0℃). No penalty is incurred when the deviation between the actual temperature and the target temperature is within the tolerance; when the deviation exceeds the tolerance, a quadratic function penalty is applied.

[0075] Objective 3 – Minimize equipment wear and tear: ,in This represents the control action (encoding of temperature, fan speed, etc.) of the i-th indoor unit at time t. This penalty is for frequent adjustment behavior to protect the compressor and motor.

[0076] Overall optimization goal:

[0077] The default values ​​for the weighting coefficients are: λ1=1.0, λ2=5.0, and λ3=0.5. Operations personnel can adjust these weights through the cloud control platform to suit different scenario preferences (e.g., prioritizing energy saving in data centers and comfort in hospitals).

[0078] 2. Hierarchical Attention Decoder Based on the fused spatiotemporal features Scheduling decisions are generated through a three-level decoder: The first layer (global scheduling layer) takes the aggregated features of all external systems as input and outputs the total load rate allocation for each external system. A Transformer encoder is used to perform cross-attention interaction on the features of all external systems, ensuring that the load rate allocation considers global coordination among the external systems. The output is the target load rate for each external system k. .

[0079] The second layer (local scheduling layer): Within each external system, the spatiotemporal fusion characteristics of all internal units under that system and the target load rate of the external units are input. It outputs the parameter adjustment values ​​for each indoor unit. Specifically, this includes: setting the temperature offset. Wind speed regulation .

[0080] The third layer (time-sequence planning layer) expands the above-mentioned immediate adjustment into a step-by-step execution sequence for the next N time steps (default N=4, i.e., the next hour), making the control action smoother and avoiding abrupt changes.

[0081] 3. Physical constraint satisfaction layer Perform constrained projection on the decoder output to ensure that the following hard constraints are met: Outdoor unit capacity constraint: The total power of all indoor units under the same outdoor unit system shall not exceed 110% of the rated capacity of the outdoor unit; Temperature change rate constraint: The set temperature change in a single adjustment shall not exceed 2℃; Minimum running time constraint: The interval between each start-stop cycle shall not be less than 3 minutes (to protect the compressor); Refrigerant balance constraint: The total cooling demand of the indoor units in the same outdoor unit system is matched with the output of the outdoor units; The constraint projection uses the gradient projection method to project decisions that do not meet the constraints to the nearest feasible region boundary.

[0082] This application also provides a multi-split air conditioning group collaborative scheduling device. Please refer to... Figure 4 , Figure 4 This is a schematic diagram of the module structure of the multi-split air conditioning group collaborative scheduling device according to an embodiment of this application. The multi-split air conditioning group collaborative scheduling device includes: The coupling relationship module 401 is used to determine the node characteristic data corresponding to each air conditioner indoor unit and the coupling relationship data between the air conditioner indoor units based on air conditioner operation data, spatial layout data and system topology data. The topology construction module 402 is used to take each indoor air conditioner unit as a graph node based on the node feature data and the coupling relationship data, and construct corresponding multi-type coupling edges according to the refrigerant system coupling relationship, horizontal spatial coupling relationship and vertical spatial coupling relationship in the coupling relationship data to obtain the topology graph of the multi-split air conditioner group. The feature extraction module 403 is used to dynamically update the edge weights of various types of coupling edges based on the topology map of the multi-split air conditioning group and the node feature data to obtain a dynamic adjacency matrix, and to perform graph attention spatial feature extraction and multi-scale temporal feature extraction based on the dynamic adjacency matrix and the node feature data to obtain spatial coupling feature data and temporal operation feature data. The gated fusion module 404 is used to perform gated fusion processing on the spatial coupling feature data and the temporal running feature data to obtain spatiotemporal fusion feature data, and to perform hierarchical scheduling decision processing based on the spatiotemporal fusion feature data to obtain candidate collaborative scheduling data. The scheduling control module 405 is used to perform constraint processing on the candidate coordinated scheduling data based on preset physical constraints to obtain a target coordinated scheduling strategy, and to perform scheduling control on the multi-split air conditioning group according to the target coordinated scheduling strategy.

[0083] The multi-split air conditioning group collaborative scheduling device provided in this application adopts the multi-split air conditioning group collaborative scheduling method in the above embodiments, which can solve the technical problem of how to improve the coordination of centralized control of building air conditioning. Compared with the prior art, the beneficial effects of the multi-split air conditioning group collaborative scheduling device provided in this application are the same as the beneficial effects of the multi-split air conditioning group collaborative scheduling method provided in the above embodiments, and other technical features in the multi-split air conditioning group collaborative scheduling device are the same as the features disclosed in the methods of the above embodiments, and will not be repeated here.

[0084] This application provides a multi-split air conditioning group collaborative scheduling device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the multi-split air conditioning group collaborative scheduling method described above.

[0085] The following is for reference. Figure 5 , Figure 5 This is a schematic diagram of the hardware operating environment involved in the multi-split air conditioning group collaborative scheduling method in the embodiments of this application. It shows a schematic diagram of the structure of the multi-split air conditioning group collaborative scheduling device suitable for implementing the embodiments of this application. Figure 5 The multi-split air conditioning group coordinated scheduling device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0086] like Figure 5As shown, the multi-split air conditioning group coordinated scheduling device may include a processing unit 1001 (e.g., a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the multi-split air conditioning group coordinated scheduling device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the multi-split air conditioning group coordination scheduling equipment to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows a multi-split air conditioning group coordination scheduling equipment with various systems, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems can be implemented alternatively.

[0087] In particular, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. When the computer program is executed by the processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0088] The multi-split air conditioning group collaborative scheduling device provided in this application, employing the multi-split air conditioning group collaborative scheduling method described in the above embodiments, can solve the technical problem of how to improve the coordination of centralized control of building air conditioning. Compared with the prior art, the beneficial effects of the multi-split air conditioning group collaborative scheduling device provided in this application are the same as those of the multi-split air conditioning group collaborative scheduling method described in the above embodiments, and other technical features in this multi-split air conditioning group collaborative scheduling device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0089] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0090] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0091] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the multi-split air conditioning group coordinated scheduling method in the above embodiments.

[0092] The aforementioned computer-readable storage medium carries one or more programs. When these programs are executed by the multi-split air conditioning group collaborative scheduling device, the multi-split air conditioning group collaborative scheduling device: determines the node characteristic data corresponding to each indoor air conditioning unit and the coupling relationship data between the indoor air conditioning units based on air conditioning operation data, spatial layout data, and system topology data; based on the node characteristic data and coupling relationship data, it treats each indoor air conditioning unit as a graph node and constructs corresponding multi-type coupling edges according to the refrigerant system coupling relationship, horizontal spatial coupling relationship, and vertical spatial coupling relationship in the coupling relationship data to obtain a multi-split air conditioning group topology graph; based on the multi-split air conditioning group topology graph... The system dynamically updates the edge weights of various types of coupled edges using node feature data to obtain a dynamic adjacency matrix. Based on the dynamic adjacency matrix and node feature data, it extracts spatial features of graph attention and multi-scale temporal features to obtain spatial coupling feature data and temporal operation feature data. It then performs gating fusion processing on the spatial coupling feature data and temporal operation feature data to obtain spatiotemporal fusion feature data. Based on this spatiotemporal fusion feature data, it performs hierarchical scheduling decision processing to obtain candidate collaborative scheduling data. Finally, it constrains the candidate collaborative scheduling data based on preset physical constraints to obtain a target collaborative scheduling strategy. The system then performs scheduling control on multi-split air conditioning groups according to this target collaborative scheduling strategy. Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof. These programming languages ​​include object-oriented programming languages—such as Java, Smalltalk, and C++—and conventional procedural programming languages—such as C or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer through any type of network—including a local area network (LAN) or a wide area network (WAN)—or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0093] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0094] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0095] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described multi-split air conditioning group coordinated scheduling method, thereby solving the technical problem of how to improve the coordination of centralized control of building air conditioning. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the multi-split air conditioning group coordinated scheduling method provided in the above embodiments, and will not be repeated here.

[0096] This application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the multi-split air conditioning group collaborative scheduling method described above.

[0097] The computer program product provided in this application can solve the technical problem of how to improve the coordination of centralized control of building air conditioning. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as the beneficial effects of the multi-split air conditioning group coordinated scheduling method provided in the above embodiments, and will not be repeated here.

[0098] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent scope of this application.

Claims

1. A method for collaborative scheduling of multi-split air conditioning groups, characterized in that, The method includes: Based on air conditioning operation data, spatial layout data, and system topology data, the node characteristic data corresponding to each air conditioning indoor unit and the coupling relationship data between air conditioning indoor units are determined. Based on the node feature data and the coupling relationship data, each indoor air conditioner unit is taken as a graph node, and corresponding multi-type coupling edges are constructed according to the refrigerant system coupling relationship, horizontal spatial coupling relationship and vertical spatial coupling relationship in the coupling relationship data to obtain the topology graph of the multi-split air conditioner group. Based on the topology of the multi-split air conditioning group and the node feature data, the edge weights of each type of coupling edge are dynamically updated to obtain a dynamic adjacency matrix. Based on the dynamic adjacency matrix and the node feature data, graph attention spatial feature extraction and multi-scale temporal feature extraction are performed to obtain spatial coupling feature data and temporal operation feature data. The spatial coupling feature data and the temporal operation feature data are subjected to gated fusion processing to obtain spatiotemporal fusion feature data, and hierarchical scheduling decision processing is performed based on the spatiotemporal fusion feature data to obtain candidate collaborative scheduling data. The candidate coordinated scheduling data is constrained based on preset physical constraints to obtain a target coordinated scheduling strategy, and the multi-split air conditioning group is scheduled and controlled according to the target coordinated scheduling strategy.

2. The method as described in claim 1, characterized in that, The steps of determining the node characteristic data corresponding to each indoor air conditioner unit and the coupling relationship data between indoor air conditioner units based on air conditioner operation data, spatial layout data, and system topology data include: The air conditioner operation data is analyzed to determine the temperature status data, operation status data and energy consumption status data of each indoor unit. Based on the temperature status data, operation status data and energy consumption status data, the corresponding operation characteristic data of each indoor unit is generated. The system topology data is parsed to determine the outdoor unit affiliation relationship, the outdoor unit system to which each indoor air conditioner unit belongs and the refrigerant association information of each indoor air conditioner unit in the corresponding outdoor unit system, and the refrigerant system coupling relationship between indoor air conditioner units is determined based on the outdoor unit system and the refrigerant association information. Based on the spatial layout data, the floor position relationship and spatial adjacency relationship of each air conditioner indoor unit are determined, and the horizontal spatial coupling relationship and vertical spatial coupling relationship between the air conditioner indoor units are determined according to the floor position relationship and the spatial adjacency relationship. The operating characteristic data is associated with the corresponding air conditioner indoor unit to obtain the node characteristic data corresponding to each air conditioner indoor unit. The coupling relationship data between the air conditioner indoor units is obtained according to the refrigerant system coupling relationship, the horizontal spatial coupling relationship and the vertical spatial coupling relationship.

3. The method as described in claim 1, characterized in that, The step of taking each indoor air conditioning unit as a graph node based on the node feature data and the coupling relationship data, and constructing corresponding multi-type coupling edges according to the refrigerant system coupling relationship, horizontal spatial coupling relationship, and vertical spatial coupling relationship in the coupling relationship data to obtain the topology graph of the multi-split air conditioning group includes: Based on the node feature data, node mapping processing is performed on each indoor air conditioner unit to determine the graph nodes that correspond one-to-one with each indoor air conditioner unit, and each graph node is associated with the corresponding node feature data to obtain a set of indoor air conditioner unit nodes. Based on the refrigerant system coupling relationship in the coupling relationship data, refrigerant association matching is performed on the graph nodes belonging to the same outdoor unit system in the set of indoor air conditioning nodes to construct refrigerant system coupling edges. Based on the horizontal spatial coupling relationship and the vertical spatial coupling relationship, horizontal spatial coupling edges and vertical spatial coupling edges are constructed respectively for the graph nodes with spatial thermal coupling relationship in the set of indoor air conditioning nodes. The refrigerant system coupling edges, the horizontal space coupling edges, and the vertical space coupling edges are labeled with edge types and their node connection relationships are integrated to obtain a set of multiple coupling edges corresponding to the set of indoor air conditioning unit nodes. Based on the set of indoor air conditioning unit nodes and the set of multiple coupling edges, the topology diagram of the multi-split air conditioning group is generated.

4. The method as described in claim 1, characterized in that, The steps of dynamically updating the edge weights of various types of coupling edges based on the topology map of the multi-split air conditioning group and the node feature data to obtain a dynamic adjacency matrix, and extracting graph attention spatial features and multi-scale temporal features based on the dynamic adjacency matrix and the node feature data to obtain spatial coupling feature data and temporal operation feature data include: Based on the edge type information, node connection relationship, and temperature status information in the node feature data of each multi-type coupling edge in the topology graph of the multi-split air conditioning group, the coupling strength corresponding to each multi-type coupling edge is dynamically characterized to obtain the updated edge weight corresponding to each multi-type coupling edge, and the dynamic adjacency matrix is ​​generated according to the updated edge weight. Based on the dynamic adjacency matrix, the set of neighboring nodes corresponding to each graph node is determined. Then, according to the node feature data, the edge type information, and the updated edge weight, graph attention calculation and feature aggregation processing are performed on the spatial influence weight between each graph node and its corresponding neighboring node to obtain the spatial coupling feature data corresponding to each air conditioner indoor unit. Based on the node feature data, a historical node feature sequence corresponding to each air conditioner indoor unit is constructed, and multi-scale causal convolution processing and scale fusion processing are performed on the historical node feature sequence to obtain the time operation feature data corresponding to each air conditioner indoor unit.

5. The method as described in claim 1, characterized in that, The steps of performing gated fusion processing on the spatial coupling feature data and the temporal operation feature data to obtain spatiotemporal fusion feature data, and performing hierarchical scheduling decision processing based on the spatiotemporal fusion feature data to obtain candidate cooperative scheduling data, include: The spatial coupling feature data and the temporal operation feature data are subjected to feature dimension matching and association encoding to obtain spatiotemporal candidate representation data, and the gating weight data corresponding to each air conditioner indoor unit is determined based on the spatiotemporal candidate representation data. Based on the gating weight data, the spatial coupling feature data and the temporal operation feature data are weighted and fused to obtain the spatiotemporal fusion feature data corresponding to each air conditioner indoor unit. According to the affiliation relationship between the air conditioner indoor unit and the outdoor unit system, the spatiotemporal fusion feature data is aggregated at the outdoor unit system level to obtain the outdoor unit system aggregated feature data. Global load allocation decisions are made based on the aggregated feature data of the outdoor system to obtain outdoor system load allocation data. In addition, indoor parameter adjustment decisions and timing execution planning are made based on the outdoor system load allocation data and the spatiotemporal fusion feature data to obtain candidate collaborative scheduling data.

6. The method as described in claim 1, characterized in that, The step of constraining the candidate coordinated scheduling data based on preset physical constraints to obtain a target coordinated scheduling strategy, and then performing scheduling control on the multi-split air conditioning group according to the target coordinated scheduling strategy, includes: The candidate collaborative scheduling data is parsed to determine the load distribution data of the external system, the parameter adjustment data of the internal system, and the time-series execution plan data. Based on the preset physical constraints, constraint verification objects are established for the load distribution data of the external system, the parameter adjustment data of the internal system, and the time-series execution plan data, respectively. Based on the constraint verification object, the feasibility of the candidate collaborative scheduling data is verified by outdoor unit capacity constraints, temperature change rate constraints, minimum running time constraints, and refrigerant balance constraints. The candidate collaborative scheduling data that does not meet the preset physical constraint conditions are subjected to constraint correction processing to obtain constrained collaborative scheduling data. Based on the constrained collaborative scheduling data, a target collaborative scheduling strategy is generated, and load allocation instructions, parameter adjustment instructions, and timing execution instructions are issued to the corresponding outdoor unit system and indoor air conditioning unit according to the target collaborative scheduling strategy, so as to perform scheduling control on the multi-split air conditioning group.

7. A multi-split air conditioning group collaborative scheduling device, characterized in that, The device includes: The coupling relationship module is used to determine the node characteristic data corresponding to each air conditioner indoor unit and the coupling relationship data between the air conditioner indoor units based on air conditioner operation data, spatial layout data and system topology data. The topology construction module is used to take each indoor air conditioner unit as a graph node based on the node feature data and the coupling relationship data, and construct corresponding multi-type coupling edges according to the refrigerant system coupling relationship, horizontal spatial coupling relationship and vertical spatial coupling relationship in the coupling relationship data to obtain the topology graph of the multi-split air conditioner group. The feature extraction module is used to dynamically update the edge weights of various types of coupling edges based on the topology map of the multi-split air conditioning group and the node feature data to obtain a dynamic adjacency matrix, and to perform graph attention spatial feature extraction and multi-scale temporal feature extraction based on the dynamic adjacency matrix and the node feature data to obtain spatial coupling feature data and temporal operation feature data. The gated fusion module is used to perform gated fusion processing on the spatial coupling feature data and the temporal running feature data to obtain spatiotemporal fusion feature data, and to perform hierarchical scheduling decision processing based on the spatiotemporal fusion feature data to obtain candidate collaborative scheduling data. The scheduling control module is used to perform constraint processing on the candidate coordinated scheduling data based on preset physical constraints to obtain a target coordinated scheduling strategy, and to perform scheduling control on the multi-split air conditioning group according to the target coordinated scheduling strategy.

8. A computer device, characterized in that, The device includes: a memory, a processor, and a multi-split air conditioning group collaborative scheduling program stored in the memory and executable on the processor, the multi-split air conditioning group collaborative scheduling program being configured to implement the steps of the multi-split air conditioning group collaborative scheduling method as described in any one of claims 1 to 6.

9. A storage medium, characterized in that, The storage medium stores a multi-split air conditioning group collaborative scheduling program, which, when executed by a processor, implements the steps of the multi-split air conditioning group collaborative scheduling method as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the steps of the multi-split air conditioning group collaborative scheduling method as described in any one of claims 1 to 6.

Images