Energy-saving control method and system for building cooling system
By dividing the open areas of large buildings into grids and analyzing their behavioral state maps, the cooling intensity and distribution of the cooling system are dynamically adjusted, resolving the contradiction between comfort and energy conservation in the open areas and achieving refined cooling control.
Patent Information
- Application Number
- CN202511145579.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-15
- Publication Date
- 2025-10-28
AI Technical Summary
In the open public areas of large buildings, existing cooling systems struggle to achieve precise temperature control, resulting in insufficient cooling in some areas and affecting user comfort. Simply pursuing energy-saving targets may lead to a discrepancy between energy efficiency and user experience.
By dividing the cooling area into grids and combining ambient temperature and user behavior data, a behavioral status map is constructed to generate a cooling energy-saving control optimization strategy. The cooling intensity and distribution are dynamically adjusted to achieve a balance between comfort and energy saving.
It achieves a fine balance between energy saving and user needs by optimizing the energy efficiency of the cooling system through fine-grained control strategies while ensuring user comfort, and dynamically adjusting the cooling intensity and distribution.
Smart Images
Figure CN120845900A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of energy-saving control technology for cooling systems, and in particular to an energy-saving control method and system for building cooling systems. Background Technology
[0002] In large buildings, centralized cooling systems typically handle the unified temperature regulation of various areas. Compared to more enclosed office or commercial areas that are easier to control locally, some open public areas inside buildings, such as rest areas, corridors, and tea rooms, are difficult to regulate precisely using a unified cooling control strategy due to their open spaces and unclear boundaries.
[0003] In such scenarios, temperature differences may exist in different local areas due to various factors such as spatial layout, personnel distribution, and ventilation disturbances. Some cooling control methods based on preset temperature control thresholds, under the guidance of energy saving, may tend to compress the overall cooling output. Although a suitable temperature may be achieved in some areas, insufficient cooling in other areas may affect users' willingness to stay and their comfort experience, thus causing a deviation between energy-saving goals and user experience.
[0004] Therefore, how to combine more granular temperature distribution information within the region with user behavior responses to establish a control strategy that can simultaneously take into account comfort and energy efficiency optimization is one of the key challenges in the current regulation of building cooling systems. Summary of the Invention
[0005] To address the aforementioned technical problems, this invention proposes an energy-saving control method and system for building cooling systems. By conducting fine-grained correlation analysis between ambient temperature and user behavior in areas with cooling demand, more suitable energy-saving regulation is achieved while considering the actual experience of users.
[0006] The first aspect of this invention provides an energy-saving control method for a building cooling system, comprising: The ambient temperature data and behavioral feature data of the target cooling area are obtained. Sliding window feature extraction is performed on the behavioral feature data to obtain multiple behavioral feature parameters corresponding to multiple sliding windows. The behavioral feature sequence of each sliding window is constructed. The target cooling area is divided into multiple local grid regions, and the first cooling sample set of each local grid region is constructed based on ambient temperature data and behavioral characteristic data. Based on the ambient temperature data, multiple candidate temperature ranges are determined. Based on the multiple candidate temperature ranges, each first cooling sample set is segmented to construct a second cooling sample set for each local grid region, which includes multiple sets of sub-cooling sample data. The behavioral feature data is discretized to generate discrete behavioral feature data of the target cooling area, and multiple candidate behavioral feature combinations are constructed for multiple behavioral feature parameters. Based on multiple second cooling sample sets, a target behavior sample set is constructed for each candidate temperature range. Based on the combination of multiple candidate behavior features and the target behavior sample set, a behavior state map is generated for each candidate temperature range. Based on multiple behavior state maps, a cooling energy-saving control optimization strategy for the target cooling area is constructed.
[0007] Preferably, for the first cooling sample set and the second cooling sample set, it further includes: The ambient temperature data and behavioral feature data are divided into grids based on multiple local grid regions to obtain local ambient temperature data and local behavioral feature data in each local grid region, and the first cooling sample set of each local grid region is constructed. Based on ambient temperature data, the target temperature control range for the target cooling area is determined. Based on a preset temperature range threshold, the target temperature control range is divided into intervals to generate multiple candidate temperature ranges for the target cooling area. Each first cooling sample set is segmented by a sliding window to obtain multiple sets of sub-cooling sample data and the sample temperature parameters corresponding to the sub-cooling sample data, thus obtaining the second cooling sample set for each local grid region.
[0008] Preferably, the process of discretizing the behavioral feature data further includes: The feature distribution range of each behavioral feature parameter is obtained by statistically analyzing multiple behavioral feature sequences. The feature distribution range of each behavioral feature parameter is then discretely segmented based on preset discrete parameters to obtain multiple discrete feature intervals for each behavioral feature parameter. Discrete feature transformation is performed on multiple behavioral feature sequences based on multiple discrete feature intervals for each behavioral feature parameter to obtain discrete behavioral feature data containing multiple discrete behavioral feature sequences. Based on multiple discrete behavioral feature sequences, multiple discrete feature values for each behavioral feature parameter are determined. The multiple discrete feature values for each behavioral feature parameter are combined to construct multiple candidate behavioral feature combinations for multiple behavioral feature parameters, wherein each candidate behavioral feature combination includes one discrete feature value of each behavioral feature parameter.
[0009] Preferably, based on combinations of multiple candidate behavioral features and a target behavioral sample set, a behavioral state map for each candidate temperature range is generated, including: For the target behavior sample set, multiple sets of target cooling sample data are selected from multiple second cooling sample sets and constructed based on the sample temperature parameters of the sub-cooling sample data. Data time series aggregation is performed on each target behavior sample set to obtain multiple sets of local cooling sample data for each target behavior sample set. The candidate behavior feature combination to which each set of target cooling sample data belongs is determined, and the behavior evolution path of each set of local cooling sample data is constructed. Multiple behavioral graph nodes are determined based on the combination of multiple candidate behavioral features, and an initial state graph is constructed for each target behavioral sample set; Multiple state transition segments are generated based on multiple behavioral evolution paths. Multiple directed edges of the initial state graph are determined based on the multiple state transition segments. Statistical analysis is performed on the multiple behavioral evolution paths to determine the state transition weight of each directed edge in the initial state graph and the state stability parameters of each behavioral graph node. Based on the behavioral feature data, the local sample weights of each group of target cooling sample data are determined. Based on the local sample weights, the multiple state transition weights and state stability parameters in the initial state map are corrected to generate the behavioral state map of each candidate temperature range.
[0010] Preferably, an energy-saving control optimization strategy for the target cooling area is constructed based on multiple behavioral state maps, including: Behavioral stability analysis and behavioral fluctuation analysis are performed on multiple behavioral state maps to generate behavioral state stability parameters and state fluctuation parameters for each behavioral state map. Based on the behavioral state stability parameters and state fluctuation parameters, the cooling adaptation index of each behavioral state map is calculated. Based on the cooling adaptation index of multiple candidate temperature ranges, the cooling energy-saving control optimization strategy for the target cooling area is constructed.
[0011] Preferably, a state-stable feature vector of the behavior state graph is constructed based on multiple modified state-stable parameters in the behavior state graph, and the entropy value of the state-stable feature vector is calculated to obtain the behavior state-stable parameters of the behavior state graph.
[0012] A second aspect of the present invention provides an energy-saving control system for a building cooling system, for implementing the above-mentioned energy-saving control method for a building cooling system, comprising: The data preprocessing module is used to acquire ambient temperature data and behavioral feature data of the target cooling area, perform sliding window feature extraction on the behavioral feature data, obtain multiple behavioral feature parameters corresponding to multiple sliding windows, and construct the behavioral feature sequence of each sliding window. The data region segmentation module is used to divide the target cooling area into multiple local grid regions, and to construct the first cooling sample set for each local grid region based on ambient temperature data and behavioral characteristic data. The data temperature segmentation module is used to determine multiple candidate temperature ranges based on ambient temperature data, segment each first cooling sample set based on multiple candidate temperature ranges, and construct a second cooling sample set for each local grid region, which includes multiple sets of sub-cooling sample data. The behavior feature combination generation module is used to perform feature discretization processing on behavior feature data, generate discrete behavior feature data of the target cooling area, and construct multiple candidate behavior feature combinations for multiple behavior feature parameters. The energy-saving control strategy generation module is used to construct a target behavior sample set for each candidate temperature range based on multiple second cooling sample sets, generate a behavior state map for each candidate temperature range based on the combination of multiple candidate behavior features and the target behavior sample set, and construct an energy-saving control optimization strategy for the target cooling area based on multiple behavior state maps.
[0013] The present invention has the following beneficial effects: This invention divides the target cooling area in a building into a grid, constructs a cooling sample set by combining ambient temperature data and behavioral characteristic data, and uses strategies such as feature discretization and state combination to explore the dynamic changes in the behavior of group users under different temperature conditions, constructing behavioral state maps for different temperature ranges. Furthermore, based on information such as state dwell and state jump in the maps that represent state changes, a cooling adaptation index is comprehensively constructed to identify the optimal temperature control range, thereby generating a cooling energy-saving control optimization strategy that can be used to dynamically adjust the cooling intensity and distribution, achieving a fine balance between energy-saving control objectives and user needs. Attached Figure Description
[0014] Figure 1 This is a flowchart illustrating an energy-saving control method for a building cooling system according to an embodiment of the present invention.
[0015] Figure 2 This is a schematic diagram of the structure of an energy-saving control system for a building cooling system according to an embodiment of the present invention. Detailed Implementation
[0016] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for explaining the invention and are not intended to limit the invention.
[0017] See Figure 1This invention provides an energy-saving control method for building cooling systems, applicable to open public areas in large buildings, such as rest areas, corridors, and tea rooms, characterized by spaciousness, blurred boundaries, and uneven distribution of human activity. This method enables in-depth modeling and dynamic control of the cooling status of the target cooling area, thereby achieving energy optimization while ensuring user comfort. Specifically, the method includes the following steps: Step S01: Obtain ambient temperature data and behavioral feature data of the target cooling area, perform sliding window feature extraction on the behavioral feature data to obtain multiple behavioral feature parameters corresponding to multiple sliding windows, and construct the behavioral feature sequence of each sliding window.
[0018] It should be noted that ambient temperature data and behavioral characteristic data can be collected by various sensor devices that are pre-deployed in the target cooling area of the building to periodically collect data related to ambient temperature and human behavior in specific areas. For example, infrared temperature sensors can be used to obtain surface temperature or air temperature at different locations in the area, human infrared sensors can be used to detect human behavior information such as entering, exiting, and moving, and pressure sensors can be used to sense information such as human stay and departure.
[0019] To capture the dynamic changes in the data, a sliding window processing mechanism is adopted. The time series data representing the behavior of user groups is divided into multiple sliding windows, and statistical features are extracted in each window, such as the average density of people, the average stay time, and the frequency of density changes. This constructs behavioral feature sequences corresponding to different sliding windows, which represent the group behavior status in the target cooling area and are used for subsequent in-depth analysis of the environment and behavior.
[0020] Step S02: Divide the target cooling area into multiple local grid regions, and construct the first cooling sample set for each local grid region based on the ambient temperature data and behavioral characteristic data.
[0021] It should be noted that the entire target area is divided into several local grid regions, such as a 2m×2m spatial grid, to determine multiple local grid regions. Local ambient temperature data and local behavioral characteristic data within each local grid region are then extracted from ambient temperature data and behavioral characteristic data. This forms local sample data for each grid region at different time periods, thus constructing the first cooling sample set for each local grid region, reflecting the user behavior response in different local areas under different temperature conditions.
[0022] Step S03: Determine multiple candidate temperature ranges based on ambient temperature data, segment each first cooling sample set based on multiple candidate temperature ranges, and construct a second cooling sample set for each local grid region, including multiple sets of sub-cooling sample data.
[0023] It should be noted that the overall temperature range corresponding to the target cooling area is determined based on the statistical distribution of ambient temperature data. Then, based on this overall temperature range, several equally divided candidate temperature range intervals are set, such as [26℃–27℃], [27℃–28℃], etc. For multiple candidate temperature ranges, the local ambient temperature data and local behavioral feature data in the first cooling sample set are processed in the same way as the behavioral feature data. The local ambient temperature data is processed by sliding window to obtain the average ambient temperature within different sliding windows. Then, according to the candidate temperature range to which they belong, the local ambient temperature data and local behavioral feature data are grouped to obtain multiple groups of sub-cooling sample data. Each group of sub-cooling sample data consists of ambient temperature and behavioral feature data corresponding to one of the sliding windows with the same timestamp. The average ambient temperature of the ambient temperature data within the window is used as the sample temperature parameter of the sub-cooling sample data. Finally, each local grid area forms several groups of sub-cooling sample data with different ambient temperatures, which can be used to analyze the differences in behavioral responses of group users under different cooling intensities.
[0024] Step S04: Perform feature discretization processing on the behavioral feature data to generate discrete behavioral feature data of the target cooling area, and construct multiple candidate behavioral feature combinations for multiple behavioral feature parameters.
[0025] It should be noted that the process of discretizing the original continuous behavioral feature parameters can be achieved by using equal-frequency segmentation, mapping the behavioral features into multiple finite discrete states. In this embodiment, the feature distribution range of each behavioral feature parameter is statistically obtained based on multiple behavioral feature sequences. Then, the feature distribution range of each behavioral feature parameter is discretized based on a preset discrete parameter, where the preset discrete parameter represents the number of states obtained from the discretization. For example, for the personnel density feature, it is discretized into five equally divided states. In this way, multiple discrete feature intervals are obtained for each behavioral feature parameter.
[0026] Then, based on multiple discrete feature intervals of each behavioral feature parameter, discrete feature transformation is performed on multiple behavioral feature sequences. In this process, quantization parameters corresponding to different discrete feature intervals can be predefined, such as quantization parameters in ascending order of 1, 2, 3, etc. In this way, multiple behavioral feature sequences are transformed into discrete behavioral feature sequences, and finally, discrete behavioral feature data containing multiple discrete behavioral feature sequences is obtained after feature discretization processing of behavioral feature data.
[0027] Furthermore, multiple discrete feature values are determined for each behavioral feature parameter based on multiple discrete behavioral feature sequences. For example, for the behavioral feature parameter of personnel density, multiple discrete feature values are obtained after discretization, representing multiple discrete states related to personnel density. Feature combinations are performed on the multiple discrete feature values of multiple behavioral feature parameters, that is, multiple discrete states. For example, for m behavioral feature parameters, each behavioral feature parameter corresponds to n discrete states. Finally, m*n candidate behavioral feature combinations for multiple behavioral feature parameters can be constructed, where each candidate behavioral feature combination includes one discrete feature value of each behavioral feature parameter. In this way, continuous behavioral feature data is quantified into multiple discrete behavioral states.
[0028] Step S05: Construct a target behavior sample set for each candidate temperature range based on multiple second cooling sample sets. Generate a behavior state map for each candidate temperature range based on the combination of multiple candidate behavior features and the target behavior sample set. Construct a cooling energy-saving control optimization strategy for the target cooling area based on multiple behavior state maps.
[0029] It should be noted that, for each candidate temperature range, data is filtered from multiple second cooling sample sets. Specifically, based on the sample temperature parameters of the sub-cooling sample data, the sub-cooling sample data that meets the candidate temperature range is marked as its corresponding target cooling sample data. In this way, multiple sets of target cooling sample data corresponding to each candidate temperature range are filtered from multiple second cooling sample sets to construct the target behavior sample set for each candidate temperature range.
[0030] Furthermore, data temporal aggregation is performed on each target behavior sample set. Specifically, target cooling sample data with temporal correlation, i.e., adjacent time windows, are aggregated together to obtain a set of local cooling sample data with temporal connection. In this way, discrete data in each target behavior sample set are aggregated to obtain multiple sets of local cooling sample data for multiple target behavior sample sets. Considering that the data in the target behavior sample set comes from different local grid regions, the aggregation can be further restricted to aggregate target cooling sample data that belong to the same local grid region and have temporal adjacency.
[0031] For the multiple candidate behavioral feature combinations constructed above, the multiple sets of target cooling sample data in the target behavioral sample set are matched with these combinations respectively. After determining the candidate behavioral feature combination to which each set of target cooling sample data belongs, the behavioral evolution path of the local cooling sample data is based on the multiple temporally changing candidate behavioral feature combinations. Finally, the behavioral evolution path of each set of local cooling sample data is constructed to indicate the characteristic behavioral evolution pattern of a group of users in a local area over a continuous period of time under the state of time change.
[0032] Then, multiple behavior graph nodes are determined based on combinations of multiple candidate behavior features to construct an initial state graph for each target behavior sample set. For the remaining structural information in the graph, micro-state segmentation is performed on multiple behavior evolution paths. Any two adjacent states in a behavior evolution path, i.e., the process of transitioning from one candidate behavior feature combination to another, are denoted as a state transition segment. After generating multiple state transition segments based on multiple behavior evolution paths, multiple directed edges of the initial state graph are determined based on these state transition segments.
[0033] After adding multiple directed edges to the initial state graph, further statistical analysis is performed on multiple behavioral evolution paths to determine the state transition weight of each directed edge in the initial state graph. Specifically, the frequency of each directed edge is first counted based on the frequency of the state transition segment corresponding to that edge in the target behavior sample set of the candidate temperature range. The number of state transition segments with state jump behaviors (i.e., the number of state transition segments with inconsistent combinations of two candidate behavioral features) is also counted. This quantity parameter is used to normalize the frequency of multiple directed edges, thus obtaining the state transition weight of each directed edge. Simultaneously, the state stability parameter of each behavioral graph node is further determined, specifically the ratio of the total number of occurrences of the candidate behavioral feature combination corresponding to the behavioral graph node in the behavioral evolution path to the number of jumps of that behavioral graph node. A higher total occurrence indicates a longer duration of group behavior remaining in that state, while fewer state jumps occurring in that state across multiple behavioral evolution paths further represent a stable or suitable state. This method quantifies the stability characteristics of candidate behavioral feature combinations.
[0034] The above method identifies multiple graph nodes, directed edges, and corresponding state transition weights representing state change characteristics in the initial state graph. It also includes state stability parameters representing the state stability characteristics of the graph nodes. However, this approach overlooks the fact that the overall usage intensity / total number of people varies across different sample periods. This might mean that some states in the graph appear due to "passive selection caused by spatial congestion" rather than "natural selection due to suitable cooling." In other words, the frequent occurrence of certain states might be because the total number of people in the target cooling area is relatively low during that period. Ignoring this phenomenon would mask the true preferences in the statistical results, leading to misjudgments of cooling suitability.
[0035] To address this situation, local sample weights are determined for each group of target cooling sample data based on behavioral feature data. Specifically, this is achieved by calculating the ratio of the total number of people in a local grid area corresponding to the target cooling sample data within the corresponding time period, and the average total number of people in multiple local grid areas within the target cooling area during that time period. This ratio is then used as the local sample weight. Subsequently, multiple state transition weights and state stability parameters in the initial state graph are corrected based on these local sample weights. Specifically, the correction of state transition weights involves determining the starting node of the directed edge corresponding to the state transition weight, and based on the local sample weight of the starting node, correcting the state transition weights of multiple directed edges belonging to the state transition states of that node. These local sample weights are then used as correction weights to achieve a weighted correction of multiple state transition weights and state stability parameters in the initial state graph, resulting in multiple corrected state transition weights and corrected state stability parameters. After correction, the weights can be naturally adjusted for states with drastic changes in pedestrian flow or low pedestrian flow, making the state change information contained in the graph more referential. The initial state graph after sample correction is recorded as the behavioral state graph of the candidate temperature range.
[0036] Each behavioral state map represents the comprehensive behavioral state of a group of users within a target cooling area within a specific temperature range. This invention performs behavioral state stability analysis and behavioral state volatility analysis on the behavioral state maps constructed for each candidate temperature range to characterize whether the behavioral performance of the group of users exhibits consistency and adaptability within a specific temperature interval.
[0037] In this embodiment, behavioral state stability analysis involves determining the corrected state stability parameters for each graph node. Based on these parameters, a state stability feature vector is constructed for each graph node. The entropy of this feature vector is then calculated and recorded as the behavioral state stability parameter of the behavioral state graph, reflecting whether user behavior tends to be concentrated and stable within the temperature range. Behavioral state volatility analysis involves determining the distance parameter between candidate behavioral feature combinations corresponding to any two graph nodes, for example, using Euclidean distance. Then, for the corrected state transition weights of multiple directed edges between two graph nodes, this distance parameter is used as the fusion weight. After determining the fusion weight of the corrected state transition weights for each directed edge in this way, the multiple corrected state transition weights are weighted and fused. The final parameter is recorded as the state volatility parameter of the behavioral state graph, reflecting the degree of drastic change in user behavior within the temperature range. Smaller state transitions indicate more natural population movement, while larger abrupt changes may indicate unsuitable temperatures within the area, causing individuals to deviate voluntarily.
[0038] Then, based on the stability parameters and fluctuation parameters of the behavioral state, the cooling adaptation index of each behavioral state map is calculated. In this embodiment, the ratio between the stability parameters and the fluctuation parameters is denoted as the cooling adaptation index. The larger the stability parameter, the more stable the overall state; the smaller the fluctuation parameter, the fewer abnormal personnel movements. Ultimately, the larger the cooling adaptation index, the more stable the behavior of the user group within the corresponding temperature zone, and the higher the degree of adaptation to the cooling needs of the user group.
[0039] By comparing the cooling adaptation index of all candidate temperature ranges, the differences in the adaptability between different temperature zones and user groups can be determined. Using the cooling adaptation index of multiple candidate temperature ranges as the data basis for cooling energy-saving control, an optimization strategy for cooling energy-saving control of the target cooling area can be constructed. In practical applications, the temperature zone with the highest adaptability can be determined as the preferred temperature range for current cooling. Then, the real-time temperature data of the target cooling area is analyzed to determine whether it deviates from the preferred temperature range. Specifically, optimization can be performed based on the overall deviation of multiple local grid areas. For example, if the temperature of a local grid area exceeding a preset proportion is lower than the preferred temperature range, it indicates that there may be over-cooling. The set temperature can be appropriately increased or the local cooling intensity reduced to achieve energy saving. Similarly, if the temperature of many local grid areas is too high, the set temperature can be appropriately decreased or the local cooling intensity increased. This avoids sacrificing user experience for excessive energy saving, achieving finer-grained energy-saving control and more appropriate energy-saving regulation while considering the actual user experience.
[0040] See Figure 2 The present invention also provides an energy-saving control system for a building cooling system, used to implement the above-described energy-saving control method for a building cooling system, the system comprising: The data preprocessing module is used to acquire ambient temperature data and behavioral feature data of the target cooling area, perform sliding window feature extraction on the behavioral feature data, obtain multiple behavioral feature parameters corresponding to multiple sliding windows, and construct the behavioral feature sequence of each sliding window. The data region segmentation module is used to divide the target cooling area into multiple local grid regions, and to construct the first cooling sample set for each local grid region based on ambient temperature data and behavioral characteristic data. The data temperature segmentation module is used to determine multiple candidate temperature ranges based on ambient temperature data, segment each first cooling sample set based on multiple candidate temperature ranges, and construct a second cooling sample set for each local grid region, which includes multiple sets of sub-cooling sample data. The behavior feature combination generation module is used to perform feature discretization processing on behavior feature data, generate discrete behavior feature data of the target cooling area, and construct multiple candidate behavior feature combinations for multiple behavior feature parameters. The energy-saving control strategy generation module is used to construct a target behavior sample set for each candidate temperature range based on multiple second cooling sample sets, generate a behavior state map for each candidate temperature range based on the combination of multiple candidate behavior features and the target behavior sample set, and construct an energy-saving control optimization strategy for the target cooling area based on multiple behavior state maps.
[0041] The above are merely specific embodiments of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art. Parts not described in detail in this specification are prior art known to those skilled in the art.
Claims
1. An energy-saving control method for a building cooling system, characterized in that, include: The ambient temperature data and behavioral feature data of the target cooling area are obtained. Sliding window feature extraction is performed on the behavioral feature data to obtain multiple behavioral feature parameters corresponding to multiple sliding windows. The behavioral feature sequence of each sliding window is constructed. The target cooling area is divided into multiple local grid regions, and the first cooling sample set of each local grid region is constructed based on ambient temperature data and behavioral characteristic data. Based on the ambient temperature data, multiple candidate temperature ranges are determined. Based on the multiple candidate temperature ranges, each first cooling sample set is segmented to construct a second cooling sample set for each local grid region, which includes multiple sets of sub-cooling sample data. The behavioral feature data is discretized to generate discrete behavioral feature data of the target cooling area, and multiple candidate behavioral feature combinations are constructed for multiple behavioral feature parameters. Based on multiple second cooling sample sets, a target behavior sample set is constructed for each candidate temperature range. Based on the combination of multiple candidate behavior features and the target behavior sample set, a behavior state map is generated for each candidate temperature range. Based on multiple behavior state maps, a cooling energy-saving control optimization strategy for the target cooling area is constructed.
2. The energy-saving control method for a building cooling system according to claim 1, characterized in that, For the first cooling sample set and the second cooling sample set, it also includes: The ambient temperature data and behavioral feature data are divided into grids based on multiple local grid regions to obtain local ambient temperature data and local behavioral feature data in each local grid region, and the first cooling sample set of each local grid region is constructed. Based on ambient temperature data, the target temperature control range for the target cooling area is determined. Based on a preset temperature range threshold, the target temperature control range is divided into intervals to generate multiple candidate temperature ranges for the target cooling area. Each first cooling sample set is segmented by a sliding window to obtain multiple sets of sub-cooling sample data and the sample temperature parameters corresponding to the sub-cooling sample data, thus obtaining the second cooling sample set for each local grid region.
3. The energy-saving control method for a building cooling system according to claim 2, characterized in that, Feature discretization of behavioral feature data also includes: The feature distribution range of each behavioral feature parameter is obtained by statistically analyzing multiple behavioral feature sequences. The feature distribution range of each behavioral feature parameter is then discretely segmented based on preset discrete parameters to obtain multiple discrete feature intervals for each behavioral feature parameter. Discrete feature transformation is performed on multiple behavioral feature sequences based on multiple discrete feature intervals for each behavioral feature parameter to obtain discrete behavioral feature data containing multiple discrete behavioral feature sequences. Based on multiple discrete behavioral feature sequences, multiple discrete feature values for each behavioral feature parameter are determined. The multiple discrete feature values for each behavioral feature parameter are combined to construct multiple candidate behavioral feature combinations for multiple behavioral feature parameters, wherein each candidate behavioral feature combination includes one discrete feature value of each behavioral feature parameter.
4. The energy-saving control method for a building cooling system according to claim 3, characterized in that, Based on combinations of multiple candidate behavioral features and a target behavioral sample set, a behavioral state map is generated for each candidate temperature range, including: For the target behavior sample set, multiple sets of target cooling sample data are selected from multiple second cooling sample sets and constructed based on the sample temperature parameters of the sub-cooling sample data. Data time series aggregation is performed on each target behavior sample set to obtain multiple sets of local cooling sample data for each target behavior sample set. The candidate behavior feature combination to which each set of target cooling sample data belongs is determined, and the behavior evolution path of each set of local cooling sample data is constructed. Multiple behavioral graph nodes are determined based on the combination of multiple candidate behavioral features, and an initial state graph is constructed for each target behavioral sample set; Multiple state transition segments are generated based on multiple behavioral evolution paths. Multiple directed edges of the initial state graph are determined based on the multiple state transition segments. Statistical analysis is performed on the multiple behavioral evolution paths to determine the state transition weight of each directed edge in the initial state graph and the state stability parameters of each behavioral graph node. Based on the behavioral feature data, the local sample weights of each group of target cooling sample data are determined. Based on the local sample weights, the multiple state transition weights and state stability parameters in the initial state map are corrected to generate the behavioral state map of each candidate temperature range.
5. The energy-saving control method for a building cooling system according to claim 4, characterized in that, Based on multiple behavioral state maps, an optimization strategy for energy-saving control of the target cooling area is constructed, including: Behavioral stability analysis and behavioral fluctuation analysis are performed on multiple behavioral state maps to generate behavioral state stability parameters and state fluctuation parameters for each behavioral state map. Based on the behavioral state stability parameters and state fluctuation parameters, the cooling adaptation index of each behavioral state map is calculated. Based on the cooling adaptation index of multiple candidate temperature ranges, the cooling energy-saving control optimization strategy for the target cooling area is constructed.
6. The energy-saving control method for a building cooling system according to claim 5, characterized in that, Based on multiple modified state stability parameters in the behavior state graph, construct the state stability feature vector of the behavior state graph, and calculate the entropy value of the state stability feature vector to obtain the behavior state stability parameters of the behavior state graph.
7. An energy-saving control system for a building cooling system, characterized in that, The system is used to implement an energy-saving control method for a building cooling system as described in any one of claims 1-6, comprising: The data preprocessing module is used to acquire ambient temperature data and behavioral feature data of the target cooling area, perform sliding window feature extraction on the behavioral feature data, obtain multiple behavioral feature parameters corresponding to multiple sliding windows, and construct the behavioral feature sequence of each sliding window. The data region segmentation module is used to divide the target cooling area into multiple local grid regions, and to construct the first cooling sample set for each local grid region based on ambient temperature data and behavioral characteristic data. The data temperature segmentation module is used to determine multiple candidate temperature ranges based on ambient temperature data, segment each first cooling sample set based on multiple candidate temperature ranges, and construct a second cooling sample set for each local grid region, which includes multiple sets of sub-cooling sample data. The behavior feature combination generation module is used to perform feature discretization processing on behavior feature data, generate discrete behavior feature data of the target cooling area, and construct multiple candidate behavior feature combinations for multiple behavior feature parameters. The energy-saving control strategy generation module is used to construct a target behavior sample set for each candidate temperature range based on multiple second cooling sample sets, generate a behavior state map for each candidate temperature range based on the combination of multiple candidate behavior features and the target behavior sample set, and construct an energy-saving control optimization strategy for the target cooling area based on multiple behavior state maps.