A method for linkage optimization of energy consumption data of multiple subsystems in a building automation system

Abstract

The present application relates to the technical field of data processing, and especially relates to a linkage optimization method for energy consumption data of multiple subsystems in a building automatic control system, which obtains historical operation mode clusters corresponding to each subsystem according to data variation characteristics in power data sequences of each subsystem in the building automatic control system, obtains historical collaborative operation modes between different subsystems according to time correlation relationships between the historical operation mode clusters of different subsystems, obtains a historical reference mode most similar to an operation mode of each subsystem in a current time window according to differences between power data of each subsystem in the current time window and each historical collaborative operation mode, and optimizes energy consumption resources of each subsystem in the building automatic control system according to a deviation degree between the power data of each subsystem in the current time window and the historical reference mode, thereby improving energy utilization efficiency of the building automatic control system.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method for the linkage optimization of energy consumption data of multiple subsystems in a building automation system. Background Technology

[0002] With the continuous development of smart communities, smart parks, and intelligent buildings, buildings typically integrate multiple subsystems, including intelligent security monitoring systems, intelligent access control systems, lighting systems, air conditioning systems, and IoT device management systems. Each subsystem continuously generates significant energy consumption during operation, collectively forming the building's overall energy consumption structure. Due to the dynamic changes in occupant activity patterns and usage demands, the load requirements and energy contributions of different subsystems vary across different time periods. Furthermore, there are often interrelationships between subsystems; for example, changes in occupant movement can affect the operation of lighting and air conditioning, while changes in the number of online devices can impact network and power supply loads. Therefore, how to conduct collaborative analysis and coordinated optimization of overall energy consumption based on the dynamic characteristics of each subsystem's operating status, thereby reducing overall energy consumption while ensuring the stable operation of various building functions, has become an important research direction in the field of building automation systems.

[0003] Existing energy optimization methods in building automation systems mostly target single systems such as air conditioning, lighting, or power equipment for energy conservation control. They typically employ preset control rules or fixed weights for energy consumption statistics and optimization decisions, focusing primarily on the energy reduction effects of local equipment or individual systems. Due to a lack of comprehensive analysis of the operational status of multiple subsystems, including security monitoring, access control, IoT devices, and building automation, as well as real-time perception of the dynamic changes in the energy consumption contribution of each subsystem, it is difficult to accurately reflect the actual impact of each subsystem on overall energy consumption under different operating scenarios. Furthermore, existing solutions generally lack a linkage optimization mechanism based on the interrelationships between multiple subsystems, which can easily lead to problems such as unreasonable selection of energy optimization targets, uneven resource allocation, and insufficient adaptability of control strategies, thereby affecting the overall building operating efficiency, energy utilization efficiency, and user experience.

[0004] Therefore, how to conduct linked analysis of energy consumption data of various subsystems in a building automation system, and then formulate better energy consumption management strategies for the building automation system to improve energy utilization efficiency, has become an urgent problem to be solved. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide a method for the linkage optimization of energy consumption data of multiple subsystems in a building automation system, in order to solve the problem of how to perform linkage analysis on the energy consumption data of each subsystem in the building automation system, and then formulate a better energy consumption management strategy for the building automation system to improve energy utilization efficiency.

[0006] This invention provides a method for the coordinated optimization of energy consumption data of multiple subsystems in a building automation system, the method comprising the following steps:

[0007] The power data of each subsystem in the building automation system is collected at each moment within a preset time period up to the current moment, and the power data sequence of each subsystem is obtained.

[0008] For any subsystem, based on the data change characteristics in the power data sequence of the subsystem, the power data sequence of the subsystem is adaptively segmented to obtain at least two operating state data segments. Based on the similarity between different operating state data segments, all operating state data segments are clustered to obtain at least two historical operating mode clusters.

[0009] Obtain the historical operation mode clusters corresponding to each subsystem. Based on the time correlation between the historical operation mode clusters of different subsystems, obtain at least two cluster combinations between different subsystems. Each cluster combination represents a historical collaborative operation mode between the subsystems.

[0010] Based on the difference between the power data of each subsystem within the preset time window up to the current moment and the power data in each cluster combination, the historical collaborative operation mode that is most similar to the operation mode of each subsystem within the preset time window is obtained and recorded as the historical reference mode.

[0011] Based on the degree of deviation between the power data of each subsystem within the preset time window and the power data in the cluster combination corresponding to the historical reference mode, the energy consumption resources of each subsystem in the building automation system are optimized.

[0012] Preferably, the step of adaptively segmenting the power data sequence of any subsystem based on the data change characteristics in the power data sequence of any subsystem to obtain at least two operating state data segments includes:

[0013] Using a sliding window of a preset size and a sliding step size, the power data sequence of any subsystem is divided, and the power data in the sliding window at the current time is removed to obtain at least two subsequences. Based on the power data difference between each pair of adjacent subsequences, the load difference degree between each pair of adjacent subsequences is obtained, and all load difference degrees are combined into a difference degree sequence.

[0014] For any load difference degree, the load difference degree is used as a candidate threshold. In the difference degree sequence, load difference degrees that are greater than or equal to the candidate threshold are obtained and recorded as target load difference degrees. The midpoint of the corresponding time period after merging the two subsequences corresponding to each target load difference degree is used as the segmentation time. According to each segmentation time, the power data in the power data sequence of any subsystem, except for the power data in the sliding window where the current time is located, is divided to obtain at least two candidate data segments.

[0015] If the number of subsequences corresponding to any candidate data segment is less than 2, the load fluctuation characteristic value of the candidate data segment is recorded as 0. If the number of subsequences corresponding to any candidate data segment is greater than or equal to 2, the average load difference between each two adjacent subsequences corresponding to the candidate data segment is calculated to obtain the load fluctuation characteristic value of the candidate data segment.

[0016] The mean value of the load fluctuation characteristic value of all candidate data segments is normalized to obtain the normalized value of the fluctuation mean. The constant 1 is subtracted from the normalized value of the fluctuation mean to obtain the first reasonable score of the candidate threshold.

[0017] The length of each candidate data segment is obtained, and the mean of the lengths of all candidate data segments is normalized to obtain the first reasonable score of the candidate threshold.

[0018] Calculate the product of the first reasonable score and the second reasonable score to obtain the segmented reasonableness evaluation value when any load difference degree is used as a candidate threshold;

[0019] Each load difference in the difference sequence is used as a candidate threshold to obtain the corresponding segment rationality evaluation value. The candidate threshold corresponding to the maximum value among all segment rationality evaluation values ​​is selected as the optimal threshold, and the candidate data segments corresponding to the optimal threshold are respectively recorded as the operating status data segments.

[0020] Preferably, the step of obtaining the load difference degree between each pair of adjacent subsequences based on the power data difference between each pair of adjacent subsequences includes:

[0021] For any two adjacent subsequences, calculate the mean of the power data in each subsequence of the two adjacent subsequences to obtain the mean load of each subsequence of the two adjacent subsequences. Normalize the absolute value of the difference between the mean loads of the two adjacent subsequences to obtain the first degree of difference between the two adjacent subsequences.

[0022] Calculate the average absolute deviation of the power data in each of the two adjacent subsequences respectively, and normalize the absolute value of the difference between the average absolute deviations of the two adjacent subsequences to obtain the second degree of difference between the two adjacent subsequences.

[0023] Calculate the mean between the first difference degree and the second difference degree to obtain the load difference degree between any two adjacent subsequences.

[0024] Preferably, the step of clustering all operating state data segments based on the similarity between different operating state data segments to obtain at least two historical operating pattern clusters includes:

[0025] For any two running status data segments, the start time of each of the two running status data segments is obtained, and the time interval between the start times of the two running status data segments is normalized to obtain the time metric distance.

[0026] Calculate the mean power data in any two operating status data segments to obtain the load mean. Normalize the absolute value of the difference between the load mean values ​​of any two operating status data segments to obtain the first degree of difference between the two operating status data segments.

[0027] Calculate the average absolute deviation of the power data in any two operating state data segments respectively, and normalize the absolute value of the difference between the average absolute deviations of the two operating state data segments to obtain the second difference degree between the two operating state data segments.

[0028] Calculate the mean between the first difference degree and the second difference degree between any two running status data segments to obtain the difference measurement distance between any two running status data segments;

[0029] Calculate the mean between the time metric distance and the difference metric distance to obtain the clustering metric distance between any two running status data segments;

[0030] Obtain the clustering metric distance between every two running status data segments. Based on the clustering metric distance between every two running status data segments, cluster all running status data segments to obtain at least two clusters, which are denoted as historical running pattern clusters.

[0031] Preferably, obtaining at least two cluster combinations among different subsystems based on the temporal correlation between the historical operation mode clusters of different subsystems includes:

[0032] For any historical operation mode cluster of any subsystem, select one historical operation mode cluster from the historical operation mode clusters of each other subsystem other than the said subsystem, and form a candidate combination with any historical operation mode cluster of the said subsystem.

[0033] Based on the temporal correlation between every two historical operation mode clusters in each candidate combination, the overall correlation degree between all historical operation mode clusters in each candidate combination is obtained, and the maximum value among all overall correlation degrees is selected as the cluster combination with any historical operation mode cluster of any subsystem as the fixed cluster.

[0034] Each subsystem is used to obtain a cluster combination with each historical operation mode cluster as the fixed cluster. Each cluster combination represents a historical collaborative operation mode between the subsystems.

[0035] Preferably, the step of obtaining the overall correlation degree among all historical operation mode clusters in each candidate combination based on the temporal correlation between every two historical operation mode clusters in each candidate combination includes:

[0036] For any two historical operation mode clusters in any candidate combination, obtain the intersection probability value of the overlapping dates and the intersection probability value of the overlapping times of the corresponding operation status data segments of the two historical operation mode clusters respectively, calculate the product of the intersection probability value of the overlapping dates and the intersection probability value of the overlapping times to obtain the correlation degree between the two historical operation mode clusters.

[0037] Obtain the correlation degree between every two historical operation mode clusters in any candidate combination, calculate the sum of all correlation degrees corresponding to any candidate combination, and obtain the overall correlation degree between all historical operation mode clusters in any candidate combination.

[0038] Preferably, the acquisition of the historical reference pattern includes:

[0039] Calculate the average power data of each subsystem within the preset time window up to the current time to obtain the average load value of each subsystem within the preset time window. Calculate the proportion of the average load value of each subsystem in the cumulative value of the average load values ​​of all subsystems to obtain the current load proportion weight of each subsystem at the current time.

[0040] For any historical collaborative operation mode corresponding to a cluster combination, calculate the average value of all power data in the historical operation mode cluster corresponding to each subsystem in the cluster combination to obtain the historical average load value of each subsystem in any historical collaborative operation mode. Calculate the proportion of the historical average load value of each subsystem in the cumulative value of the historical average load values ​​of all subsystems to obtain the historical load proportion weight of each subsystem in any historical collaborative operation mode.

[0041] Calculate the absolute value of the difference between the historical load percentage weight and the current load percentage weight of each subsystem, and sum all the absolute values ​​of the difference to obtain the structural difference index between any historical collaborative operation mode and the operation mode of each subsystem within the preset time window;

[0042] Obtain the structural difference index corresponding to each historical collaborative operation mode, and select the historical collaborative operation mode with the minimum value among all structural difference indices as the historical reference mode.

[0043] Preferably, the optimization of energy consumption resources of each subsystem in the building automation system based on the deviation between the power data of each subsystem within the preset time window and the power data in the cluster combination corresponding to the historical reference mode includes:

[0044] For any subsystem, the target difference is obtained by subtracting the historical load percentage weight of the subsystem under the historical reference mode from the current load percentage weight of the subsystem at the current moment. The target difference is then used as the independent variable of the natural exponential function to obtain the relative deviation of the subsystem at the current moment.

[0045] Calculate the product between the current load percentage of any subsystem at the current moment and the relative deviation degree to obtain the load optimization priority coefficient of any subsystem at the current moment;

[0046] Obtain the load optimization priority coefficient of each subsystem at the current moment, and generate differentiated optimization strategies based on the load optimization priority coefficient of each subsystem at the current moment to achieve energy resource optimization of each subsystem in the building automation system.

[0047] The beneficial effects of the embodiments of the present invention compared with the prior art are as follows:

[0048] This invention first adaptively segments the power data sequences of each subsystem in a building automation system (BAS) based on the data change characteristics, obtaining at least two operating state data segments. Then, based on the similarity between different operating state data segments, all operating state data segments are clustered to obtain at least two historical operating mode clusters for each subsystem. Subsequently, through the temporal correlation between the historical operating mode clusters of different subsystems, a multi-subsystem collaborative operating mode is constructed. Based on the degree of difference between the current load ratio structure of each subsystem and the historical collaborative operating mode, the historical reference mode closest to the current operating scenario is determined. Furthermore, combining the current load ratio structure of each subsystem and its deviation from the historical reference mode, the load optimization priority coefficient of each subsystem is calculated. Based on the load optimization priority coefficient, a differentiated linkage optimization strategy is generated to optimize the energy consumption resources of each subsystem in the BAS. This improves the targeting and adaptive capability of the overall energy consumption optimization of the BAS, reducing overall energy consumption and improving energy utilization efficiency while ensuring the stable operation of the BAS. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 This is a flowchart of a method for the linkage optimization of energy consumption data of multiple subsystems in a building automation system, provided in Embodiment 1 of the present invention. Detailed Implementation

[0051] Embodiments of this disclosure are described in detail below, with examples of these embodiments illustrated in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this disclosure, and should not be construed as limiting it.

[0052] It should be noted that the terms "first," "second," etc., used in this disclosure and the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure.

[0053] To illustrate the technical solution of the present invention, specific embodiments are described below.

[0054] See Figure 1 This is a flowchart of a method for the linkage optimization of energy consumption data of multiple subsystems in a building automation system, as provided in Embodiment 1 of the present invention. Figure 1 As shown, the method may include:

[0055] Step S101: Collect power data of each subsystem in the building automation system at each moment within a preset time period up to the current moment to obtain the power data sequence of each subsystem.

[0056] Most existing energy optimization methods in building automation systems focus on energy-saving control of single systems, typically employing preset control rules or fixed weights for energy consumption statistics and optimization decisions. These methods primarily address the energy reduction effects of local equipment or individual systems. However, in actual operation, different subsystems experience varying degrees of load fluctuations due to changes in human activity, environment, and user demand. For example, increased personnel traffic often leads to a simultaneous increase in the operating load of access control, lighting, and air conditioning systems. Conversely, during low-load periods at night, the operating status of each subsystem changes accordingly. Therefore, the same subsystem may operate in different modes at different times. Analyzing only the complete historical load sequence of a single system is susceptible to the influence of mixed operating scenarios. It lacks real-time perception of the dynamic changes in the energy consumption contribution of each subsystem, making it difficult to accurately reflect the actual impact of each subsystem on overall energy consumption under different operating scenarios. This can easily lead to problems such as unreasonable selection of energy optimization targets, uneven resource allocation, and insufficient adaptability of control strategies, ultimately affecting the overall building operating efficiency, energy utilization efficiency, and user experience.

[0057] Therefore, in this embodiment of the invention, by performing a linkage analysis on the operating load of each subsystem, the historical collaborative operation mode between multiple subsystems is determined, and the historical collaborative operation mode closest to the current scenario is determined as a historical reference. Then, based on the difference between the current operating load of each subsystem and the historical reference, the load optimization priority coefficient of each subsystem is calculated, and a differentiated linkage optimization strategy is generated accordingly. The operating parameters of the subsystems are dynamically adjusted to achieve accurate identification and priority optimization of subsystems with abnormal energy consumption, improve the pertinence and adaptability of the overall building energy consumption optimization, reduce overall energy consumption while ensuring the stable operation of the building, and improve energy utilization efficiency.

[0058] First, data collection is performed. Through the building automation system, power data for each subsystem within the building is acquired at each moment within the past month, up to the current time. This results in a power data sequence for each subsystem, reflecting its operational load. Subsystems include, but are not limited to, security monitoring systems, access control systems, lighting systems, HVAC systems, and IoT device management platforms. The power data is collected once per second, with no fixed limit; implementers can set this according to specific scenarios. The collected power data from each subsystem undergoes time synchronization processing and is sorted and stored according to a unified time benchmark, forming a historical load database covering multiple subsystems. This provides a data foundation for subsequent energy consumption optimization.

[0059] Step S102: For any subsystem, based on the data change characteristics in the power data sequence of the subsystem, adaptive segmentation is performed on the power data sequence of the subsystem to obtain at least two operating state data segments. Based on the similarity between different operating state data segments, all operating state data segments are clustered to obtain at least two historical operating mode clusters.

[0060] The same subsystem may correspond to different operating modes at different time periods. In order to obtain the operating load characteristics of each subsystem at different time periods, so as to identify the energy consumption performance of each subsystem under different operating scenarios and further optimize the subsystems with relatively abnormal energy consumption, it is necessary to obtain the historical operating modes of each subsystem. Taking subsystem A as an example, the implementation method is as follows:

[0061] (1) Based on the data change characteristics in the power data sequence of subsystem A, the power data sequence of subsystem A is adaptively segmented to obtain at least two operating state data segments.

[0062] First, a sliding window is established to initially segment the power data sequence of subsystem A. The length of the sliding window is set to 10 minutes, and the sliding step size is half the length of the sliding window. There is no limit to this; the implementer can set it according to the specific scenario. When more precise identification of load changes is required, the window length can be shortened. Using the sliding window and sliding step size, the power data sequence of subsystem A is divided, and the power data in the sliding window at the current moment is removed, resulting in at least two subsequences. Since the data in the sliding window at the current moment is the data to be analyzed, the power data in it may be too large, so it is not used as the object of analysis of the operating status for the time being. That is, the power data in the sliding window at the current moment is removed. In other words, the subsequences obtained above do not include the power data in the sliding window at the current moment. Subsequently, only the power data before the sliding window at the current moment is analyzed to obtain the historical operating pattern, which is convenient for comparing the current operating status with the historical operating pattern, thereby obtaining the energy consumption status of each subsystem at the current moment.

[0063] Taking the i-th subsequence and the (i+1)-th subsequence as examples, the mean of the power data in the i-th subsequence and the (i+1)-th subsequence are calculated respectively to obtain the load mean of the i-th subsequence and the (i+1)-th subsequence. The absolute value of the difference between the load mean of the i-th subsequence and the (i+1)-th subsequence is normalized to obtain the first degree of difference between the i-th subsequence and the (i+1)-th subsequence.

[0064] Calculate the average absolute deviation of the power data in the i-th subsequence and the (i+1)-th subsequence respectively, and normalize the absolute value of the difference between the average absolute deviations of the i-th subsequence and the (i+1)-th subsequence to obtain the second degree of difference between the i-th subsequence and the (i+1)-th subsequence.

[0065] Calculate the mean between the first difference degree and the second difference degree to obtain the load difference degree between the i-th subsequence and the (i+1)-th subsequence;

[0066] In one embodiment, the formula for calculating the load difference between the i-th subsequence and the (i+1)-th subsequence is:

[0067]

[0068] In the formula, This represents the load difference between the i-th subsequence and the (i+1)-th subsequence. This represents the mean load of the i-th subsequence. This represents the load mean of the (i+1)th subsequence. This represents the mean absolute deviation of the i-th subsequence. This represents the average absolute deviation of the (i+1)th subsequence. Represents the absolute value symbol. Represents the normalization function;

[0069] It should be noted that, and Let represent the overall load of the i-th subsequence and the (i+1)-th subsequence, respectively. and Let represent the load change characteristics of the i-th subsequence and the (i+1)-th subsequence, respectively. The larger, or The larger the value, the more likely the i-th subsequence and the (i+1)-th subsequence are to represent two different operating modes of subsystem A, and thus... The larger.

[0070] Similarly, the load difference degree between every two adjacent subsequences is obtained, and all load difference degrees are combined into a difference degree sequence. Further, each load difference degree is used sequentially as a candidate threshold to segment the load time series data, specifically:

[0071] Taking the Tth load difference as an example, the Tth load difference is used as a candidate threshold. In the difference sequence, load differences greater than or equal to the candidate threshold are obtained and recorded as target load differences. The midpoint of the time period corresponding to the two subsequences of each target load difference is taken as the segmentation time. According to each segmentation time, the power data in the power data sequence of any subsystem, except for the power data in the sliding window where the current time is located, is divided to obtain at least two candidate data segments. This realizes the merging operation of the two subsequences corresponding to load differences less than the candidate threshold.

[0072] If the number of subsequences corresponding to any candidate data segment is less than 2, then the load fluctuation characteristic value of the candidate data segment is recorded as 0. If the number of subsequences corresponding to any candidate data segment is greater than or equal to 2, that is, the candidate data segment consists of at least two subsequences, then the average load difference between each two adjacent subsequences corresponding to the candidate data segment is calculated to obtain the load fluctuation characteristic value of the candidate data segment.

[0073] The mean value of the load fluctuation characteristic value of all candidate data segments is normalized to obtain the normalized value of the fluctuation mean. The constant 1 is subtracted from the normalized value of the fluctuation mean to obtain the first reasonable score of the candidate threshold.

[0074] The length of each candidate data segment is obtained, and the mean of the lengths of all candidate data segments is normalized to obtain the first reasonable score of the candidate threshold.

[0075] Calculate the product between the first reasonable score and the second reasonable score to obtain the segmented reasonableness evaluation value when the Tth load difference degree is used as a candidate threshold.

[0076] In one embodiment, the formula for calculating the segmentation rationality evaluation value when the Tth load difference degree is used as a candidate threshold is:

[0077]

[0078] In the formula, This represents the segmentation rationality evaluation value when the T-th load difference is used as a candidate threshold. This represents the average length of all candidate data segments obtained when the T-th load variability is used as the candidate threshold. This represents the number of all candidate data segments obtained when the T-th load difference is used as the candidate threshold. This represents the load fluctuation characteristic value of the s-th candidate data segment obtained when the T-th load difference is used as the candidate threshold. This represents the normalization function.

[0079] It should be noted that, The larger the value, the lower the probability of over-segmentation, and the more reasonable the threshold selection. The larger; The larger the threshold, the smaller the load fluctuation within the candidate data segment, meaning a higher probability that subsystem A will not experience a state transition within the candidate data segment. This indicates a more reasonable threshold selection. The larger.

[0080] Similarly, each load difference in the difference sequence is used as a candidate threshold to obtain the corresponding segment rationality evaluation value. Among all segment rationality evaluation values, the candidate threshold corresponding to the maximum value is selected as the optimal threshold, and the candidate data segments corresponding to the optimal threshold are respectively recorded as the running status data segments.

[0081] The above steps adaptively identify the load state switching position based on the load changes of subsystem A, and complete the adaptive segmentation processing of the historical load data of subsystem A, avoiding the phenomenon of over-segmentation or under-segmentation of load data due to manually setting fixed thresholds.

[0082] (2) Based on the similarity between different running status data segments, cluster all running status data segments to obtain at least two historical running mode clusters.

[0083] Since different operating status data segments may correspond to operating states under different time scenarios, it is necessary to further identify the similarity relationships between each operating status data segment in order to extract the historical operating pattern characteristics of subsystem A and obtain at least two historical operating mode clusters of subsystem A. Specifically: considering that the operating load of each subsystem in the building automation system usually has obvious time period patterns and the operating scenarios corresponding to different time periods are different, a time distance constraint is introduced when analyzing similarity. That is, for any two operating status data segments, the start time of the two operating status data segments is obtained respectively, and the time interval between the start times of the two operating status data segments is normalized to obtain the time metric distance.

[0084] Similar to the calculation method of the load difference between the i-th subsequence and the (i+1)-th subsequence, the difference measurement distance between any two operating state data segments is calculated. That is, the mean value of the power data in any two operating state data segments is calculated to obtain the load mean value. The absolute value of the difference between the load mean values ​​between any two operating state data segments is normalized to obtain the first difference between any two operating state data segments.

[0085] Calculate the average absolute deviation of the power data in any two operating state data segments respectively, and normalize the absolute value of the difference between the average absolute deviations of the two operating state data segments to obtain the second difference degree between the two operating state data segments.

[0086] Calculate the mean between the first difference degree and the second difference degree between any two running status data segments to obtain the difference measurement distance between any two running status data segments;

[0087] Calculate the mean between the time metric distance and the difference metric distance to obtain the clustering metric distance between any two running status data segments.

[0088] In one embodiment, taking the a-th running state data segment and the b-th running state data segment as an example, the formula for calculating the clustering metric distance between the a-th running state data segment and the b-th running state data segment is as follows:

[0089]

[0090]

[0091] In the formula, This represents the clustering metric distance between the a-th and b-th running status data segments. This represents the time interval between the start times of the a-th and b-th running status data segments. This represents the distance metric indicating the difference between the a-th and b-th running status data segments. This represents the average load of the a-th operating status data segment. This represents the average load of the b-th operating status data segment. This represents the average absolute deviation of the a-th running status data segment. This represents the average absolute deviation of the b-th running status data segment. Represents the absolute value symbol. This represents the normalization function.

[0092] Similarly, the clustering metric distance between every two running status data segments is obtained. Based on the clustering metric distance between every two running status data segments, a time distance constraint is introduced into the clustering metric distance, which allows the clustering results to take into account both the running load characteristics and the time scene characteristics, thereby improving the accuracy of historical running pattern recognition.

[0093] Furthermore, the HDBSCAN clustering algorithm is used to cluster all operational status data segments, resulting in at least two clusters. The operational status data segments within each cluster exhibit similar load variation patterns and temporal distribution characteristics, which can be used to characterize the typical operational modes of subsystem A under different operating scenarios, providing a foundation for subsequent multi-subsystem correlation analysis and coordinated energy consumption optimization. For ease of subsequent description, each cluster is denoted as the historical operational mode cluster. The HDBSCAN clustering algorithm is existing technology and will not be elaborated upon here.

[0094] Step S103: Obtain the historical operation mode clusters corresponding to each subsystem. Based on the time correlation between the historical operation mode clusters of different subsystems, obtain at least two cluster combinations between different subsystems. Each cluster combination represents a historical collaborative operation mode between each subsystem.

[0095] Following the method used to obtain the historical operation mode clusters of subsystem A, the historical operation mode clusters of each subsystem in the building automation system are obtained. Since subsystems such as access control, security monitoring, lighting, and air conditioning do not operate independently but are collectively affected by factors such as human activity patterns, environmental changes, and usage demands, there are usually certain collaborative operation relationships between different subsystems. For example, when people enter the building in large numbers, the load on the access control system increases, and the loads on the lighting and air conditioning systems also rise simultaneously; in low-traffic scenarios at night, the loads on the lighting and access control systems decrease, while the operating rate of the security system relatively increases. Therefore, to identify the collaborative operation modes between different subsystems, it is necessary to further establish the correlation between the historical operation mode clusters of each subsystem to obtain multiple historical collaborative operation modes between the subsystems. The implementation process is as follows:

[0096] Taking the u-th historical operation pattern cluster of subsystem A as an example, and using it as a "fixed value", we randomly select one historical operation pattern cluster from the historical operation pattern clusters of each other subsystem besides subsystem A, and form a candidate combination with the u-th historical operation pattern cluster of subsystem A. Assume there are x subsystems in total, and each subsystem has If there are *u* historical operation mode clusters, then when the *u*th historical operation mode cluster of subsystem A is a "constant value", there are a total of *u* historical operation mode clusters. A number of candidate combinations.

[0097] Taking the k-th candidate combination as an example, based on the temporal correlation between every two historical operation pattern clusters in the k-th candidate combination, the overall correlation degree among all historical operation pattern clusters in the k-th candidate combination is obtained. Specifically:

[0098] Assuming the k-th candidate combination includes the u-th historical operating pattern cluster of subsystem A and the v-th historical operating pattern cluster of subsystem B, we obtain the intersection probability values ​​of the overlapping dates and overlapping times (overlapping times on the same day) of the operating status data segments corresponding to the u-th and v-th historical operating pattern clusters, respectively. We then calculate the product of these two intersection probability values ​​to obtain the correlation degree between the u-th and v-th historical operating pattern clusters of subsystem A and subsystem B, denoted as […]. ,Right now In the formula, This represents the probability value of the intersection of overlapping dates. Specifically, it is calculated by dividing the number of intersections of all dates by the number of unions of all dates, based on the dates corresponding to each running status data segment of the u-th and v-th historical running pattern clusters. The intersection probability value representing time overlap is calculated by dividing the number of intersections of all time points (i.e., minutes and seconds, ignoring dates) by the number of unions of all time points, based on the time points (i.e., seconds, minutes, and seconds) of each power data point corresponding to the u-th and v-th historical operating mode clusters. When two clusters simultaneously have high date co-occurrence probabilities and time overlap probabilities, it indicates that they are more likely to be driven by the same operating scenario, thus exhibiting a stronger cooperative operating relationship. The larger.

[0099] Similarly, calculate the correlation between every two historical operation pattern clusters in the k-th candidate combination, and calculate the sum of all correlations corresponding to the k-th candidate combination to obtain the overall correlation between all historical operation pattern clusters in the k-th candidate combination.

[0100] Similarly, calculate the overall correlation degree of all candidate combinations when the u-th historical operation mode cluster of subsystem A is a "fixed value". Select the maximum value among all overall correlation degrees as the cluster combination with the u-th historical operation mode cluster of subsystem A as the fixed cluster. This represents the historical collaborative operation mode of each subsystem of the building automation system under a certain typical operation scenario, which is also the load distribution structure of each subsystem of the building automation system under a certain typical operation scenario. Similarly, obtain the cluster combination with each historical operation mode cluster of each subsystem as the fixed cluster (removing duplicates).

[0101] Step S104: Based on the difference between the power data of each subsystem within the preset time window up to the current moment and the power data in each cluster combination, obtain the historical collaborative operation mode that is most similar to the operation mode of each subsystem within the preset time window, and record it as the historical reference mode.

[0102] After obtaining the historical collaborative operation modes between each subsystem through the above steps, a mapping relationship between the historical collaborative operation modes and the current operation status is constructed. The historical collaborative operation mode that is most similar to the current operation status of each subsystem is obtained to reflect the current load deviation of each subsystem. This enables accurate identification and priority optimization of subsystems with abnormal energy consumption, improves the pertinence and adaptability of building overall energy consumption optimization, reduces overall energy consumption while ensuring stable building operation, and improves energy utilization efficiency.

[0103] The specific method for obtaining the historical collaborative operation mode that is most similar to the current operating state of each subsystem is as follows:

[0104] Calculate the average power data of each subsystem within a preset time window up to the current time to obtain the average load value of each subsystem within the preset time window. The length of the preset time window is the same as the length of the sliding window in step S102, which is 10 minutes in this embodiment. Calculate the proportion of the average load value of each subsystem in the cumulative value of the average load values ​​of all subsystems to obtain the current load proportion weight of each subsystem at the current time.

[0105] Taking subsystem A as an example, the formula for calculating the current load percentage weight of subsystem A at the current moment is:

[0106]

[0107] In the formula, This indicates the current load percentage weight of subsystem A at the current moment. This represents the average load value of subsystem A over the past 10 minutes up to the current time, and m represents the number of all subsystems in the building automation system.

[0108] For any historical collaborative operation mode corresponding to a cluster combination, calculate the average value of all power data in the historical operation mode cluster corresponding to each subsystem in the cluster combination to obtain the historical average load value of each subsystem in any historical collaborative operation mode. Calculate the proportion of the historical average load value of each subsystem in the cumulative value of the historical average load values ​​of all subsystems to obtain the historical load proportion weight of each subsystem in any historical collaborative operation mode.

[0109] Calculate the absolute value of the difference between the historical load percentage weight and the current load percentage weight of each subsystem, and sum all the absolute values ​​of the difference to obtain the structural difference index between any historical collaborative operation mode and the operation mode of each subsystem within the preset time window. The smaller the structural difference index, the closer the current load distribution structure between each subsystem is to any historical collaborative operation mode; conversely, the larger the index, the greater the difference between the two.

[0110] Obtain the structural difference index corresponding to each historical collaborative operation mode. Select the historical collaborative operation mode with the minimum value from all structural difference indices as the one most similar to the current operating state of each subsystem. For ease of subsequent description, this is designated as the historical reference mode. This mode is closest to the current load structure and therefore can serve as a reference for the reasonable load allocation relationship of each subsystem under the current operating scenario.

[0111] Step S105: Optimize the energy consumption resources of each subsystem in the building automation system based on the degree of deviation between the power data of each subsystem within the preset time window and the power data in the cluster combination corresponding to the historical reference mode.

[0112] After obtaining the historical reference model through the above steps, the load optimization priority coefficient of each subsystem at the current moment is obtained based on the degree of deviation between the power data of each subsystem within 10 minutes up to the current moment and the historical reference model, so as to realize the collaborative optimization and dynamic control of energy consumption data of multiple subsystems in the building.

[0113] Taking subsystem A as an example, the steps to obtain the load optimization priority coefficient of subsystem A at the current moment are as follows:

[0114] Subtract the historical load percentage weight of subsystem A under the historical reference mode from the current load percentage weight of subsystem A at the current moment to obtain the target difference. Use the target difference as the independent variable of the natural exponential function to obtain the relative deviation of subsystem A at the current moment.

[0115] The load optimization priority coefficient of subsystem A at the current moment is obtained by multiplying the current load percentage of subsystem A at the current moment with the relative deviation degree.

[0116] In one implementation, the formula for calculating the load optimization priority coefficient of subsystem A at the current moment is:

[0117]

[0118] In the formula, This represents the load optimization priority coefficient of subsystem A at the current moment. This indicates the current load percentage weight of subsystem A at the current moment. represents the historical load percentage weight of subsystem A under the historical reference mode, and e represents the natural index.

[0119] It should be noted that, The larger the value, the greater the contribution of subsystem A to the overall building automation system's energy consumption within the last 10 minutes up to the current moment. In other words, the higher the load proportion of subsystem A, the greater its impact on the overall energy consumption of the building automation system. Therefore, optimizing the energy consumption of subsystem A should be prioritized. The larger; The larger the value, the higher the current load percentage of subsystem A is compared to its historical load percentage under the historical reference mode. In this case, priority should be given to optimizing the energy consumption of subsystem A. The larger.

[0120] Similarly, by obtaining the load optimization priority coefficient of each subsystem in the building automation system at the current moment, the higher the priority coefficient, the higher the energy consumption of the subsystem at the current moment, and the more significant the load deviation compared to the historical reasonable operating state. Subsystems with lower priority coefficients can maintain their current operating state based on actual conditions, thereby achieving coordinated optimization and dynamic control of energy consumption data across multiple building subsystems. Therefore, differentiated optimization strategies can be generated based on the load optimization priority coefficients of each subsystem to optimize energy consumption resources in the building automation system. For example, when the load optimization priority coefficient of the air conditioning system is high, the temperature setpoint, fan operating frequency, or zone operation strategy can be adjusted appropriately to reduce its operating power consumption; when the load optimization priority coefficient of the lighting system is high, the brightness or number of lights turned on in non-critical areas can be dynamically adjusted; when the load optimization priority coefficient of the security monitoring system or IoT device management platform is high, the device operating frequency, data acquisition frequency, or resource allocation strategy can be adjusted according to current needs. For subsystems with lower optimization priority, the current operating state can be maintained or the control intensity reduced based on actual conditions, thereby avoiding impact on normal operation.

[0121] After optimization, new power data for each subsystem is continuously collected, and the load share analysis, historical collaborative operation mode matching, and load optimization priority coefficient calculation processes are re-executed. For example, if the load share of the air conditioning system gradually decreases from 60% to 45% corresponding to the historical reference mode, while the load shares of the lighting system and access control system become closer to the historical collaborative operation mode, or if the difference between the weights of each load share and the historical load share weights is less than a preset threshold, such as 0.1, then the current optimization strategy is effective; otherwise, the optimization strategy is readjusted. This dynamic feedback optimization mechanism gradually brings the load structure of each subsystem closer to the historically efficient collaborative operation mode, thereby achieving continuous optimization of the building's overall energy consumption.

[0122] It is worth noting that the focus of this invention is on how to perform a coordinated analysis of the energy consumption of each subsystem in a building automation system, obtain a historical reference pattern that is most similar to the current operating state, and then obtain the load optimization priority coefficient of each subsystem based on the degree of deviation between the current load distribution and the historical reference pattern. Developing differentiated optimization strategies based on the load optimization priority coefficient to achieve energy resource optimization is existing technology and will not be elaborated upon here.

[0123] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for coordinated optimization of energy consumption data of multiple subsystems in a building automation system, characterized in that, The method for coordinated optimization of energy consumption data of multiple subsystems in a building automation system includes: The power data of each subsystem in the building automation system is collected at each moment within a preset time period up to the current moment, and the power data sequence of each subsystem is obtained. For any subsystem, based on the data change characteristics in the power data sequence of the subsystem, the power data sequence of the subsystem is adaptively segmented to obtain at least two operating state data segments. Based on the similarity between different operating state data segments, all operating state data segments are clustered to obtain at least two historical operating mode clusters. Obtain the historical operation mode clusters corresponding to each subsystem. Based on the time correlation between the historical operation mode clusters of different subsystems, obtain at least two cluster combinations between different subsystems. Each cluster combination represents a historical collaborative operation mode between the subsystems. Based on the difference between the power data of each subsystem within the preset time window up to the current moment and the power data in each cluster combination, the historical collaborative operation mode that is most similar to the operation mode of each subsystem within the preset time window is obtained and recorded as the historical reference mode. Based on the degree of deviation between the power data of each subsystem within the preset time window and the power data in the cluster combination corresponding to the historical reference mode, the energy consumption resources of each subsystem in the building automation system are optimized.

2. The method for coordinated optimization of energy consumption data of multiple subsystems in a building automation system according to claim 1, characterized in that, The step involves adaptively segmenting the power data sequence of any subsystem based on the data change characteristics in the power data sequence of any subsystem to obtain at least two operating state data segments, including: Using a sliding window of a preset size and a sliding step size, the power data sequence of any subsystem is divided, and the power data in the sliding window at the current time is removed to obtain at least two subsequences. Based on the power data difference between each pair of adjacent subsequences, the load difference degree between each pair of adjacent subsequences is obtained, and all load difference degrees are combined into a difference degree sequence. For any load difference degree, the load difference degree is used as a candidate threshold. In the difference degree sequence, the load difference degree that is greater than or equal to the candidate threshold is obtained and recorded as the target load difference degree. The midpoint of the corresponding time period after merging the two subsequences corresponding to each target load difference degree is used as the segmentation time. According to each segmentation time, the power data in the power data sequence of any subsystem, except for the power data in the sliding window where the current time is located, is divided to obtain at least two candidate data segments. If the number of subsequences corresponding to any candidate data segment is less than 2, the load fluctuation characteristic value of the candidate data segment is recorded as 0. If the number of subsequences corresponding to any candidate data segment is greater than or equal to 2, the average load difference between each two adjacent subsequences corresponding to the candidate data segment is calculated to obtain the load fluctuation characteristic value of the candidate data segment. The mean value of the load fluctuation characteristic value of all candidate data segments is normalized to obtain the normalized value of the fluctuation mean. The constant 1 is subtracted from the normalized value of the fluctuation mean to obtain the first reasonable score of the candidate threshold. The length of each candidate data segment is obtained, and the mean of the lengths of all candidate data segments is normalized to obtain the first reasonable score of the candidate threshold. Calculate the product of the first reasonable score and the second reasonable score to obtain the segmented reasonableness evaluation value when any load difference degree is used as a candidate threshold; Each load difference in the difference sequence is used as a candidate threshold to obtain the corresponding segment rationality evaluation value. The candidate threshold corresponding to the maximum value among all segment rationality evaluation values ​​is selected as the optimal threshold, and the candidate data segments corresponding to the optimal threshold are respectively recorded as the operating status data segments.

3. The method for coordinated optimization of energy consumption data of multiple subsystems in a building automation system according to claim 2, characterized in that, The step of obtaining the load difference degree between each pair of adjacent subsequences based on the power data difference between each pair of adjacent subsequences includes: For any two adjacent subsequences, calculate the mean of the power data in each subsequence of the two adjacent subsequences to obtain the mean load of each subsequence of the two adjacent subsequences. Normalize the absolute value of the difference between the mean loads of the two adjacent subsequences to obtain the first degree of difference between the two adjacent subsequences. Calculate the average absolute deviation of the power data in each of the two adjacent subsequences respectively, and normalize the absolute value of the difference between the average absolute deviations of the two adjacent subsequences to obtain the second degree of difference between the two adjacent subsequences. Calculate the mean between the first difference degree and the second difference degree to obtain the load difference degree between any two adjacent subsequences.

4. The method for coordinated optimization of energy consumption data of multiple subsystems in a building automation system according to claim 1, characterized in that, The process involves clustering all operational status data segments based on their similarity to obtain at least two historical operational pattern clusters, including: For any two running status data segments, the start time of each of the two running status data segments is obtained, and the time interval between the start times of the two running status data segments is normalized to obtain the time metric distance. Calculate the mean value of the power data in any two operating status data segments to obtain the load mean value. Normalize the absolute value of the difference between the load mean values ​​of any two operating status data segments to obtain the first degree of difference between the two operating status data segments. Calculate the average absolute deviation of the power data in any two operating state data segments respectively, and normalize the absolute value of the difference between the average absolute deviations of the two operating state data segments to obtain the second difference degree between the two operating state data segments. Calculate the mean between the first difference degree and the second difference degree between any two running status data segments to obtain the difference measurement distance between any two running status data segments; Calculate the mean between the time metric distance and the difference metric distance to obtain the clustering metric distance between any two running status data segments; Obtain the clustering metric distance between every two running status data segments. Based on the clustering metric distance between every two running status data segments, cluster all running status data segments to obtain at least two clusters, which are denoted as historical running pattern clusters.

5. The method for coordinated optimization of energy consumption data of multiple subsystems in a building automation system according to claim 1, characterized in that, The step of obtaining at least two cluster combinations among different subsystems based on the temporal correlation between clusters according to the historical operation modes of different subsystems includes: For any historical operation mode cluster of any subsystem, select one historical operation mode cluster from the historical operation mode clusters of each other subsystem other than the said subsystem, and form a candidate combination with any historical operation mode cluster of the said subsystem. Based on the temporal correlation between every two historical operation mode clusters in each candidate combination, the overall correlation degree between all historical operation mode clusters in each candidate combination is obtained, and the maximum value among all overall correlation degrees is selected as the cluster combination with any historical operation mode cluster of any subsystem as the fixed cluster. Each subsystem is used to obtain a cluster combination with each historical operation mode cluster as the fixed cluster. Each cluster combination represents a historical collaborative operation mode between the subsystems.

6. The method for coordinated optimization of energy consumption data of multiple subsystems in a building automation system according to claim 5, characterized in that, The step of obtaining the overall correlation degree among all historical operation pattern clusters in each candidate combination based on the temporal correlation between every two historical operation pattern clusters in each candidate combination includes: For any two historical operation mode clusters in any candidate combination, obtain the intersection probability value of the overlapping dates and the intersection probability value of the overlapping times of the corresponding operation status data segments of the two historical operation mode clusters respectively, calculate the product of the intersection probability value of the overlapping dates and the intersection probability value of the overlapping times, and obtain the correlation degree between the two historical operation mode clusters. Obtain the correlation degree between every two historical operation mode clusters in any candidate combination, calculate the sum of all correlation degrees corresponding to any candidate combination, and obtain the overall correlation degree between all historical operation mode clusters in any candidate combination.

7. The method for coordinated optimization of energy consumption data of multiple subsystems in a building automation system according to claim 1, characterized in that, The acquisition of the historical reference pattern includes: Calculate the average power data of each subsystem within the preset time window up to the current time to obtain the average load value of each subsystem within the preset time window. Calculate the proportion of the average load value of each subsystem in the cumulative value of the average load values ​​of all subsystems to obtain the current load proportion weight of each subsystem at the current time. For any historical collaborative operation mode corresponding to a cluster combination, calculate the average value of all power data in the historical operation mode cluster corresponding to each subsystem in the cluster combination to obtain the historical average load value of each subsystem in any historical collaborative operation mode. Calculate the proportion of the historical average load value of each subsystem in the cumulative value of the historical average load values ​​of all subsystems to obtain the historical load proportion weight of each subsystem in any historical collaborative operation mode. Calculate the absolute value of the difference between the historical load percentage weight and the current load percentage weight of each subsystem, and sum all the absolute values ​​of the difference to obtain the structural difference index between any historical collaborative operation mode and the operation mode of each subsystem within the preset time window; Obtain the structural difference index corresponding to each historical collaborative operation mode, and select the historical collaborative operation mode with the minimum value among all structural difference indices as the historical reference mode.

8. The method for coordinated optimization of energy consumption data of multiple subsystems in a building automation system according to claim 7, characterized in that, The optimization of energy consumption resources of each subsystem in the building automation system is based on the deviation between the power data of each subsystem within the preset time window and the power data in the cluster combination corresponding to the historical reference mode, including: For any subsystem, the target difference is obtained by subtracting the historical load percentage weight of the subsystem under the historical reference mode from the current load percentage weight of the subsystem at the current moment. The target difference is then used as the independent variable of the natural exponential function to obtain the relative deviation of the subsystem at the current moment. Calculate the product between the current load percentage of any subsystem at the current moment and the relative deviation degree to obtain the load optimization priority coefficient of any subsystem at the current moment; Obtain the load optimization priority coefficient of each subsystem at the current moment, and generate differentiated optimization strategies based on the load optimization priority coefficient of each subsystem at the current moment to achieve energy resource optimization of each subsystem in the building automation system.

Images