Intelligent building load directrix generation method, device, equipment, medium and product

Abstract

The invention discloses an intelligent building load directrix generation method, device and equipment, a medium and a product, and belongs to the technical field of power grid dispatching and control. The method comprises the following steps: carrying out load curve feature extraction on historical load data of each intelligent building in an intelligent building cluster to obtain a feature vector; clustering the intelligent building load curves to generate an intelligent building cluster typical load characteristic curve and various types of intelligent building load shape curves; according to the system response demand curve and the typical load characteristic curve of the intelligent building cluster, generating an ideal load directrix of the intelligent building cluster; with the ideal load directrix as a target, clustering response capability of the intelligent building is aggregated, and a clustering response capability aggregation result is obtained; generating an intelligent building cluster load directrix; and decomposing the intelligent building cluster load directrix into various types of intelligent building load directrix. According to the embodiment of the invention, the accuracy of the generated load directrix of the intelligent building can be improved, and the efficiency of the intelligent building participating in demand side response is improved.

Description

Technical Field

[0001] This application relates to the field of power grid dispatching and control technology, and in particular to a method, apparatus, equipment, medium and product for generating intelligent building load baselines. Background Technology

[0002] In the context of the energy internet, smart buildings, as integrated energy hubs combining photovoltaics, flexible loads, and vehicle-to-grid interaction, possess considerable regulation potential and require demand-side response to achieve deep coupling between building energy efficiency and grid dispatch. Existing demand-side response mechanisms mainly include two forms: price-based (such as time-of-use pricing and real-time pricing) and incentive-based (such as interruptible load management and ancillary service plans).

[0003] Among them, the load profile method, as an incentive-based response mechanism, optimizes power grid operation by guiding user load curves to approximate the ideal system curve. However, existing load profile generation methods are mainly based on system economy and security objectives during the formulation process, without fully considering the regulation characteristics of the load itself. This may lead to insufficient load regulation capacity in actual response, resulting in the failure to achieve the expected response rate and weakening the effectiveness of the load profile. Summary of the Invention

[0004] The purpose of this application is to provide a method, apparatus, device, medium, and product for generating load profiles for intelligent buildings, which can effectively improve the accuracy of the generated load profiles for intelligent buildings, thereby enhancing the efficiency of intelligent buildings in participating in demand-side response.

[0005] To achieve the above objectives, a first aspect of this application provides a method for generating intelligent building load guidelines, comprising: The historical load data of each smart building in the smart building cluster are used to extract load curve features, resulting in feature vectors including ramp features and edge point features. Based on the feature vector, clustering algorithms are used to cluster the load curves of smart buildings to generate typical load characteristic curves of smart building clusters and load shape curves of various types of smart buildings. Based on the system response demand curve and the typical load characteristic curve of the intelligent building cluster, an ideal load guideline for the intelligent building cluster is generated. Using the ideal load guideline as the target, the cluster response capabilities of intelligent buildings are aggregated through optimization models to obtain the cluster response capability aggregation results; Based on the aggregated results of the cluster response capabilities, a load baseline for the intelligent building cluster is generated. With the goal of minimizing the total adjustment of each smart building, the load baseline of the smart building cluster is decomposed into load baselines for each type of smart building.

[0006] Compared with existing technologies, the intelligent building load profile generation method provided in this application has the following advantages: By extracting core features of load curves such as ramps and edge points and combining them with clustering algorithms, it accurately depicts the overall load characteristics and internal type differences of intelligent building clusters, solving the problem that existing technologies cannot reflect the diversity of cluster loads; by combining system response requirements with typical characteristics of intelligent building clusters to generate ideal load profiles for intelligent building clusters, and aggregating the actual response capabilities of the clusters to generate feasible intelligent building cluster load profiles, it fully considers load regulation characteristics, avoids insufficient response rate due to the profile deviating from the actual response capability, and improves the effectiveness of intelligent building cluster load profiles; finally, with the goal of minimizing the total regulation, the cluster profile is decomposed into load profiles for each type of intelligent building, reducing the response cost of a single intelligent building, achieving a multi-objective balance between system requirements, cluster capabilities, and the response cost of a single intelligent building, and ultimately significantly improving the efficiency and economy of intelligent buildings participating in demand-side response.

[0007] In some embodiments, the load curve feature extraction includes: The load curve of the intelligent building is normalized to obtain the shape of the load curve. Extract ramp features, defined as events in which the degree of change in the load curve exceeds a first preset threshold within a set time window; Extract edge point features, defined as an event where the slope of the load curve changes beyond a second preset threshold at a certain moment; Extract routine load characteristics, including daily maximum load, daily peak-to-valley difference, daily peak-to-valley difference rate, and daily load factor.

[0008] In some embodiments, the clustering algorithm employs the k-means++ algorithm and determines the number of cluster categories using the silhouette coefficient method; the typical load characteristic curve of the intelligent building cluster is obtained by setting the number of cluster categories to 1.

[0009] In some embodiments, generating the ideal load profile for a smart building cluster specifically involves superimposing the system response demand curve with the typical load characteristic curve of the smart building cluster.

[0010] In some embodiments, the optimization model for the response capability of the aggregated intelligent building cluster includes: Objective function: Minimize the deviation between the ideal load baseline and the actual total cluster load; Constraints include response capacity constraints, maximum response duration constraints, response frequency constraints, maximum number of responses constraints, and response speed constraints.

[0011] In some embodiments, generating the intelligent building cluster load baseline specifically involves: generating a response curve based on the cluster response capability aggregation result using a robust optimization model, and overlaying it with the typical load characteristic curve of the intelligent building cluster.

[0012] To achieve the above objectives, a second aspect of this application provides an intelligent building load baseline generation device, the device comprising: The extraction module is used to extract load curve features from the historical load data of each smart building in the smart building cluster, and obtain feature vectors including ramp features and edge point features. The clustering module is used to cluster the load curves of smart buildings based on the feature vectors using a clustering algorithm, and generate typical load characteristic curves of smart building clusters and load shape curves of various types of smart buildings. The generation module is used to generate an ideal load guideline for the intelligent building cluster based on the system response demand curve and the typical load characteristic curve of the intelligent building cluster. The aggregation module is used to aggregate the response capabilities of intelligent building clusters by optimizing the model, with the ideal load guideline as the target, to obtain the aggregated cluster response capabilities result; The processing module is used to generate a load baseline for the intelligent building cluster based on the aggregation results of the cluster response capabilities. The decomposition module is used to decompose the load baseline of the intelligent building cluster into load baselines of various types of intelligent buildings with the goal of minimizing the total adjustment of each intelligent building.

[0013] To achieve the above objectives, a third aspect of this application provides an electronic device, the electronic device including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the method described in the first aspect.

[0014] To achieve the above objectives, a fourth aspect of the present application provides a computer-readable storage medium comprising a stored computer program, wherein the computer program, when executed, controls the device containing the computer-readable storage medium to perform the method described in the first aspect.

[0015] To achieve the above objectives, a fifth aspect of the present application provides a computer program product, which includes a computer program or computer instructions, wherein the computer program or computer instructions, when executed by a processor, implement the method described in the first aspect. Attached Figure Description

[0016] Figure 1 This is a flowchart of a method for generating intelligent building load guidelines provided in an embodiment of this application; Figure 2 It is a demand response curve for new energy consumption scenarios; Figure 3 This is a clustering result diagram of the load curves; Figure 4 These are typical curves for various load types; Figure 5 This is a typical load curve for a cluster of intelligent buildings; Figure 6 It is an ideal load profile for intelligent building clusters; Figure 7 This is a graph showing the aggregated response results of a smart building cluster under the scenario of peak reduction and valley filling; Figure 8 This is a comparison chart of the response aggregation results for cluster 1; Figure 9 This is a schematic diagram of the decomposition of the response quantities of a smart building cluster; Figure 10 It is a cluster load baseline diagram of smart buildings in peak shaving and valley filling scenarios; Figure 11 This is a schematic diagram of the load guideline decomposition results for a smart building cluster. Figure 12 It is a load baseline diagram for various types of intelligent buildings; Figure 13 This is a comparison chart of the accessibility results of the load baseline of intelligent building clusters; Figure 14 These are load baseline diagrams for various intelligent buildings without considering their control characteristics. Figure 15 This is a schematic diagram of the intelligent building load guideline generation device provided in the embodiments of this application; Figure 16 This is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0018] In the description of this application, it should be understood that the terms "center", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0019] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0020] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0021] Against the backdrop of the development of the energy internet, smart buildings have gradually evolved into comprehensive energy hubs integrating photovoltaic systems, flexible load clusters, and vehicle-to-grid (V2G) interaction facilities, possessing considerable regulation potential. There is an urgent need to achieve deep coupling between building energy efficiency management and grid dispatch commands through demand-side response mechanisms. Current research has relatively complete models of smart buildings themselves and their internal operating modes, and has also analyzed their interaction with the grid from a market perspective. However, a complete theory guiding smart building responses under various typical demand-side regulation scenarios has not yet been formed. This guidance requires first clarifying the demand-side response forms and then aggregating and evaluating the response capabilities of smart buildings and their clusters.

[0022] Current demand-side response mechanisms are mainly divided into two categories: price-based and incentive-based. Price-based mechanisms rely on dynamic pricing (such as time-of-use pricing and real-time pricing) to guide users to optimize load patterns, forming price-driven regulation capabilities. Incentive-based mechanisms establish a collaborative mechanism between the power grid and users through contractual agreements (such as interruptible load management and ancillary service plans), and conduct economic settlement based on response deviations. Among these, incentive-based response based on load profiles is an important form—the load profile is an ideal load curve derived based on system economy and safety. Users participate in the response by closely approximating this curve. Its advantage over baseline-based response is that it can consider the relationship between the load profile curve and the overall system curve, better meeting the overall control needs of smart buildings. Smart buildings can optimize overall electricity consumption and energy storage plans through control terminals, rather than just adjusting individual devices.

[0023] However, existing load profile technologies have significant shortcomings: First, the formulation process does not take into account the load's own regulation characteristics (such as response capability and adjustment speed), which may lead to the failure to achieve the expected response rate due to insufficient load regulation capability, thus weakening the effectiveness of the profile; second, it is difficult to fully reflect the overall load curve characteristics and internal diversity of intelligent building clusters; and third, it lacks a load profile decomposition mechanism for various types of intelligent buildings within the cluster, which increases the response cost of intelligent buildings.

[0024] Based on this, embodiments of this application provide a method, apparatus, device, medium, and product for generating load guidelines for intelligent buildings, which can effectively improve the accuracy of the generated load guidelines for intelligent buildings.

[0025] Please see Figure 1 , Figure 1 This is an optional flowchart of the intelligent building load guideline generation method provided in the embodiments of this application. Figure 1 The method may include, but is not limited to, steps S101 to S106.

[0026] Step S101: Extract load curve features from the historical load data of each smart building in the smart building cluster to obtain feature vectors including ramp features and edge point features; Step S102: Based on the feature vector, clustering algorithm is used to cluster the load curves of smart buildings to generate typical load characteristic curves of smart building clusters and load shape curves of various types of smart buildings. Step S103: Generate the ideal load guideline for the intelligent building cluster based on the system response demand curve and the typical load characteristic curve of the intelligent building cluster. Step S104: Using the ideal load baseline as the target, the cluster response capability of intelligent buildings is aggregated through optimization model to obtain the cluster response capability aggregation result; Step S105: Based on the cluster response capability aggregation results, generate a smart building cluster load baseline; Step S106: With the goal of minimizing the total adjustment of each smart building, the load baseline of the smart building cluster is decomposed into load baselines for each type of smart building.

[0027] Steps S101 to S106 of this embodiment accurately characterize the overall load characteristics and internal type differences of the intelligent building cluster by extracting core features of the load curve such as ramp and edge points and combining them with clustering algorithms, thus solving the problem that existing technologies cannot reflect the diversity of cluster loads. By combining system response requirements with typical characteristics of the intelligent building cluster to generate an ideal load guideline for the intelligent building cluster, and aggregating the actual response capabilities of the cluster to generate a feasible load guideline for the intelligent building cluster, the load regulation characteristics are fully considered, avoiding insufficient response rate due to the guideline deviating from the actual response capability, and improving the effectiveness of the intelligent building cluster load guideline. Finally, with the goal of minimizing the total regulation, the cluster guideline is decomposed into load guidelines for each type of intelligent building, reducing the response cost of a single intelligent building, achieving a multi-objective balance between system requirements, cluster capabilities, and the response cost of a single intelligent building, and ultimately significantly improving the efficiency and economy of intelligent buildings participating in demand-side response.

[0028] First, we will introduce the characteristics and basic interaction modes of demand-side response in smart buildings. Smart buildings encompass diverse load resources such as uncontrollable electrical loads, smart air conditioners, smart home appliances, and electric vehicles. The main characteristic of their demand-side response lies in their overall control capability—unlike the individual adjustments of traditional adjustable loads (such as responding solely through air conditioning temperature control). Smart buildings can rely on control terminals that integrate data acquisition, information forwarding, and strategy execution functions to coordinate diverse internal load resources, achieving full-process management including electricity consumption forecasting and analysis, response capability assessment, electricity consumption planning, and real-time decision-making. They possess the ability to monitor, assess, coordinate, and control internal loads, which is something that general adjustable loads lack when performing demand-side response.

[0029] Because the response capabilities of a single smart building are limited, it needs to participate in demand-side response in the form of a smart building cluster. The basic interaction process between smart buildings and the power grid is as follows: First, the control terminal of a single smart building collects response information of various types of adjustable loads within it and statistically analyzes the overall response characteristics of the smart building; second, the cluster operator aggregates the response characteristics of each smart building to form a smart building cluster response aggregate, which interacts with the power trading center; third, the power trading center and the cluster operator conduct two-way information communication to formulate a load baseline at the cluster level; subsequently, the cluster operator decomposes the cluster load baseline into load baselines for each smart building and distributes them to the control terminals of each smart building; finally, the control terminal formulates operation plans for various types of loads within the smart building based on the load baseline of the smart building, so that the building load curve conforms to the load baseline of the smart building.

[0030] In step S101 of some embodiments, the intelligent building cluster can be a collection of multiple intelligent buildings with load regulation capabilities, which can collaboratively participate in demand-side response. Historical load data can be the power consumption-time data of each intelligent building over a past period. The load curve is a curve plotted with time as the horizontal axis and power consumption as the vertical axis, reflecting the changing patterns of building electricity consumption.

[0031] The core difference between a load baseline and a load guideline lies in the latter's utilization of the inherent curve characteristics of the load itself. Therefore, before establishing a load guideline, it is necessary to first clarify the load curve characteristics and response capabilities of individual smart buildings and clusters. Based on this, feature extraction is first performed on the load curves of smart buildings.

[0032] First, to focus on the trend of load variation over time (excluding the interference of total electricity consumption differences on the curve shape), the shape of the smart building load curve is defined as a normalized load curve. It is obtained by normalizing the original load curve through the total daily electricity consumption of the smart building load, and the calculation formula is as follows: (1); In formula (1), This is the original load curve for a smart building (reflecting the actual power consumption at different times). This represents the total daily electricity consumption (total daily power consumption) of the smart building. After normalization, It only reflects the time-varying pattern of load and is not affected by the total electricity consumption.

[0033] Based on the normalized load curve, this application selects ramp features and edge point features as core extraction indicators to capture the dynamic change pattern of the load curve.

[0034] Ramp-up characteristic: Used to identify scenarios where power in the load curve rises rapidly (uphill) or falls rapidly (downhill). It is defined as follows: within a set time window, if the power change in the load curve exceeds a first preset threshold... If the condition is met, then the window is determined to contain climbing characteristics; otherwise, it is not. The calculation formula is as follows: (2); In formula (2), =1 indicates that there is a climbing characteristic in this time window. =0 indicates no climbing characteristics.

[0035] Edge point feature: Used to locate critical moments in the trend reversal of the load curve (such as from rising to falling, or from falling to rising). It is defined as follows: if the difference or sum of the slopes calculated from the load value at a certain moment and the load values ​​at two adjacent moments exceeds a second preset threshold... If the value is zero, then that moment is an edge point. The calculation formula is as follows: (3); In formula (3), , These are the slopes of the load curves at this moment, compared to the previous moment, and the next moment (reflecting the direction and speed of load change). =1 indicates that the point is an edge point at that moment. =0 represents a non-edge point.

[0036] The above two features allow for the complete extraction of the dynamic trend and inflection points of the load curve. In some embodiments, if the load types of smart buildings differ significantly (e.g., office buildings and commercial buildings), or if there is a substantial difference in load power, the following additional conventional load features can be used to describe the load characteristics from a macroscopic perspective: Maximum daily load: The maximum load power at all times during the day, reflecting the peak electricity consumption during the day. The formula is... (4); Daily peak-to-valley difference: The difference between the daily maximum load and the daily minimum load, reflecting the intraday load fluctuation range, calculated using the formula: (5); Daily peak-to-valley ratio: The ratio of the daily peak-to-valley difference to the daily maximum load, quantifying the degree of load fluctuation in a proportional form. The formula is: (6); Daily load factor: The ratio of the daily average load to the daily maximum load, reflecting load utilization efficiency (the higher the value, the more stable the load). The formula is as follows: (7); After extracting the load curve features, the core and supplementary features extracted above are integrated based on the historical load data of the smart building to construct the feature vector of the smart building. The mathematical expression is as follows: (8) In formula (8), For smart buildings The eigenvectors of have a total of . A characteristic of the load curve. For example, =[Number of climbing features, Number of edge point features], where daily maximum load, daily peak-to-valley difference, daily peak-to-valley difference rate, and daily load rate can be supplemented into the feature vector according to actual application needs. middle.

[0037] It should be noted that if the overall load trend of each smart building in the smart building cluster is consistent (the peak and valley periods are basically the same), dimensionality reduction can be achieved by counting the number of climbing features and edge point features throughout the entire time period. If the feature extraction time span is long and the number of features is large, multiple time windows can be set, and the number of the two types of core features can be counted in each window to optimize the feature vector dimension.

[0038] In step S102 of some embodiments, the clustering algorithm can be an algorithm that groups samples based on data similarity, which is used here to classify smart building types according to the similarity of load characteristics; the typical load characteristic curve can be the average load curve representing the overall electricity consumption pattern of the entire smart building cluster after clustering (reflecting the commonality of the cluster); the load shape curve of each type of smart building can be the average load curve representing the electricity consumption pattern of that type of building (such as office buildings, commercial buildings) in each clustering result (reflecting the characteristics of the type).

[0039] This application employs the k-means++ algorithm to cluster the load curves of each smart building within a smart building cluster based on the constructed feature vectors of the smart buildings. Because k-means++ selects more dispersed initial cluster centers probabilistically, it effectively reduces the dependence of the traditional k-means algorithm on initial centers, thus improving clustering accuracy. The silhouette coefficient method is used to determine the optimal number of clusters. The silhouette coefficient ranges from -1 to 1; the closer it is to 1, the better the clustering effect (high similarity among samples of the same class and large differences among samples of different classes). The calculation formula is as follows: (9); In formula (9), For the sample The average distance to other samples in the same cluster. For the sample The average distance to other samples outside the same cluster. In this embodiment of the application, the center curve of each cluster after clustering is defined as the intelligent building load shape curve, which represents the typical load shape of a certain type of intelligent building (such as office buildings or commercial buildings).

[0040] In addition to multi-cluster clustering, single-cluster clustering is also required to generate a curve reflecting the load pattern of the entire intelligent building cluster. During single-cluster clustering, the number of cluster categories must be forcibly set to 1. This single-cluster clustering curve ignores the type differences between buildings within the intelligent building cluster, integrates the load curve characteristics of all intelligent buildings, and obtains a curve that represents the overall electricity consumption pattern of the cluster; this curve is defined as the typical load characteristic curve of the intelligent building cluster (denoted as...). This curve will be directly used in the subsequent generation of the load baseline for intelligent building clusters.

[0041] In step S103 of some embodiments, the system response demand curve can be the ideal response power-time curve that the power grid requires the load cluster to achieve based on supply and demand balance and safety / economic objectives; the ideal load baseline can be a cluster-level target curve that is initially generated by combining system demand and cluster characteristics, without considering actual response capabilities.

[0042] Intelligent buildings and their associated response clusters, as complex entities with diverse response characteristics, exhibit significant differences in the response features of their constituent entities, making simple equivalence through generators or energy storage devices difficult. Traditional demand-side response capability aggregation methods aim to maximize the total response volume by describing the feasible domain of the response cluster. While this maximizes the representation of response capability, it fails to address specific response scenarios. The essence of demand-side response aggregation is to dimensionality-reducedly map the complete feasible domain of adjustable equipment to the feasible domain of external connection lines. To achieve lossless projection, mathematical tools such as Fourier-Motzkin elimination are needed to transform all internal variable constraints into connection-level constraints. However, such methods introduce numerous new constraints into the dimensionality-reduction model, leading to a dramatic increase in mathematical complexity. This makes them difficult to implement when modeling complex adjustable resources and large-scale resource clusters, thus limiting their practical application. Furthermore, commonly used lossy dimensionality reduction methods inevitably incur information loss during the dimensionality reduction process, resulting in discrepancies between the actual and aggregated response capabilities.

[0043] For smart building aggregators, minimizing the information loss from aggregation dimensionality reduction when assessing cluster response capabilities allows for more accurate control of adjustable resources, thereby improving demand-side response transaction revenue. To this end, this embodiment first generates an ideal load profile for the smart building cluster, then embeds this ideal load profile as a target into the aggregation process, generating the actual load profile through the cluster response capabilities obtained from targeted aggregation.

[0044] First, the power dispatching department needs to predict the different response scenarios the system may face the following day, including peak shaving, valley filling, renewable energy consumption, renewable energy fluctuation suppression, and emergency backup. Specifically, external features such as Long Short-Term Memory (LSTM) neural network algorithms can be used to generate time-series response demand curves based on historical demand-side scenarios. For example, the response demand curve for the photovoltaic consumption scenario could be as follows: Figure 2 As shown; alternatively, historical demands from different response scenarios can be superimposed to generate a response demand curve for a composite scenario (the specific prediction models for response demand curves have been extensively studied and will not be elaborated upon in this application). In general, this application can generate the required time-series response demand curve based on historical load data of intelligent building clusters using a Long Short-Term Memory (LSTM) neural network algorithm.

[0045] Based on this, by superimposing the response demand curve for a specific scenario with the aforementioned typical load characteristic curve of the intelligent building cluster (the normalized values ​​need to be restored to actual values), the ideal load guideline for the intelligent building cluster can be obtained, and its mathematical expression is: (10); In formula (10), for Total response demand at any given time; For intelligent building clusters The percentage of electricity used in all participating response clusters; express Smart Building Cluster Typical load characteristic curves of intelligent building clusters; Represents intelligent building clusters Under ideal conditions, participate in the total load curve after the response; for Smart Building Cluster Ideal load guideline value; For intelligent building clusters The total load. The essence of generating the ideal load profile for a smart building cluster is to superimpose the response quantity that needs to be scheduled in a certain scenario with the original load operation curve, thereby obtaining an ideal load profile that combines the inherent load pattern of the smart building cluster with the system response requirements.

[0046] In step S104 of some embodiments, the optimization model can be a mathematical model (such as a linear programming model or a particle swarm optimization model) used to solve for the optimal solution. Here, the objective is to make the response capability of the intelligent building cluster conform to the ideal guideline, and the constraints are the response capacity constraint, the maximum response duration constraint, the response frequency constraint, the maximum number of responses constraint, and the response speed constraint for each intelligent building. The aggregated result of the cluster response capability can be a comprehensive parameter (reflecting the upper limit of the actual response that the cluster can achieve) such as the maximum achievable response power range of the cluster as a whole, calculated through the optimization model.

[0047] After obtaining the ideal load profile, in order to minimize information loss during the aggregation projection process, this application embeds the load profile into the aggregation process of the intelligent building cluster response capability. The core is to optimize the aggregation to make the external response curve obtained as close as possible to the ideal load profile. The specific aggregation model is as follows: Objective function: (11); In formula (11), for Total load of smart building clusters at any given time; for Smart Building Cluster Ideal load guideline value; for The original load of each smart building at any given time. For the total electricity consumption of smart buildings to be aggregated, the core of the objective function is to minimize the deviation between the response demand and the actual response, and to ensure that the aggregation result conforms to the control target of the ideal load guideline.

[0048] The constraints include response capacity constraints, maximum response duration constraints, response frequency constraints, maximum number of responses constraints, and response speed constraints.

[0049] Response capacity constraints: (12); (13); In formulas (12) and (13), for Total load of smart building clusters at any given time; Indicating intelligent buildings Response volume; This indicates the number of smart buildings in the smart building cluster; Represents the response status. Representatives did not participate in the response. Representatives participated in the response. Represents the minimum response size for a smart building. This represents the maximum response capacity of a smart building. This response capacity constraint is used to limit the boundary of a single building's adaptability.

[0050] Maximum response duration constraint: (14); In formula (14), This represents the maximum duration of a single response from a smart building. The maximum response duration constraint describes the maximum duration for which a smart building can respond after receiving a response curve.

[0051] Response frequency constraint: (15); In formula (15), This is the cumulative response time since the last response. The minimum time interval for response. The response frequency constraint describes the time interval between receiving the next response curve and executing it in a smart building after receiving the first response curve.

[0052] Maximum number of responses constraint: (16); In formula (16), This represents the maximum number of times a response curve can be executed within a day. The maximum number of responses constraint describes the maximum number of times a smart building can execute a response curve in a single day.

[0053] Response speed constraints: (17); In formula (17), To reduce the rate of load reduction, The rate at which the load is increased. Response speed constraints are used to describe how quickly a smart building reaches its response capacity after receiving a curve.

[0054] This aggregation model, based on satisfying the various response constraints of intelligent buildings (i.e., corresponding to formulas (12)-(17)), optimizes the adjustment of each intelligent building to minimize the deviation between the total cluster adjustment and the ideal load guideline. The sum of the optimal solutions obtained from solving the objective function constitutes... The upper and lower bounds of this curve reflect the maximum responsiveness that a cluster of intelligent buildings can achieve under the guidance of the ideal load baseline (i.e., the result of cluster response capability aggregation). This process is essentially a directional aggregation projection—projecting the original feasible domain of the system onto the ideal load baseline to achieve precise aggregation of response capabilities.

[0055] In step S105 of some embodiments, after obtaining the feasible region of the response capability of the intelligent building cluster, in order to cope with the randomness in the response process, a robust optimization model considering the randomness of the response is constructed to generate the final response curve for the intelligent building cluster. The mathematical model is as follows: (18); In formula (18), for Total load of smart building clusters at any given time; This refers to the system's response requirements in a specific response scenario. For intelligent building clusters Response volume; The number of smart building clusters participating in the response; This represents the minimum response time for a cluster of intelligent buildings in this scenario. This represents the maximum response volume of the intelligent building cluster in this scenario. Let be the response randomness coefficient. Considering the worst-case response, we ensure that the response distribution requirements are still met even with the minimum response coefficient, which can ultimately be transformed into a deterministic problem to solve.

[0056] Based on this, the load guideline of the intelligent building cluster can be obtained by superimposing the response curves of each intelligent building cluster obtained by solving them with the aforementioned typical load characteristic curves of the intelligent building cluster. The mathematical expression is as follows: (19); In formula (19), for Smart Building Cluster The load baseline value; Represents intelligent building clusters The total load curve after participating in the response; for Total response demand at any given time; express Smart Building Cluster Typical load characteristic curves of intelligent building clusters; For intelligent building clusters The total load. By fully incorporating the control characteristics of intelligent buildings (response capacity, speed, duration, etc.), it is ensured that the intelligent building cluster can effectively track the load baseline of the intelligent building cluster within its own response capability range.

[0057] In step S106 of some embodiments, the total adjustment amount can be the sum of the absolute values ​​of the load power that needs to be adjusted up / down throughout the entire time period, with a single smart building as the benchmark (the smaller the total adjustment amount, the lower the building response cost).

[0058] If the load profile of a smart building cluster is directly used as the basis for the demand-side response of each smart building, it will be impossible to reduce adjustment costs by utilizing the diverse load curve shapes of different buildings. Therefore, the cluster load profile needs to be decomposed according to the types of load curves, and the decomposition result is called the smart building load profile. The purpose of the decomposition is to minimize the total response of a single smart building, while satisfying the constraint that the sum of the responses of each smart building equals the total response of the smart building cluster. Since the load profile only contains curve feature information, the decomposition process needs to combine the electricity consumption proportion of each type of load curve in the cluster to weight the decomposition result.

[0059] The mathematical model for the decomposition of clustered directrixes is as follows: (20); In formula (20), The number of categories of load curve shapes for smart buildings in the cluster; For the first The percentage of electricity used in smart buildings; for Intelligent building load guideline The value at time; for The load shape curve of intelligent buildings in The value at time; for The load guideline value of the intelligent building cluster at any given time. By solving formula (20), the optimal solution of the load guideline of each type of building with the minimum response cost under the premise of satisfying the overall shape of the intelligent building cluster guideline can be obtained. That is, the first The load baseline value of a smart building at time t. Through the above decomposition, each smart building can select the appropriate load baseline for execution based on its own load curve type, thereby optimizing response costs.

[0060] Under the load baseline mechanism, the interaction process between smart buildings and the power grid is as follows: The power trading center or dispatch center generates an ideal load baseline by combining historical load demand, renewable energy output curves, weather data, etc., and provides it to the cluster operator; The cluster operator collects the response characteristic information of smart buildings, generates an aggregated result curve of the smart building cluster response capability based on the ideal load baseline, and sends it back to the power trading center or dispatch center; The power trading center or dispatch center corrects the ideal load baseline based on the aggregated result, generates the cluster load baseline, and sends it to the cluster operator; The cluster operator decomposes the cluster load baseline to obtain the load baseline of each smart building and sends it to its control terminal. The control terminal then adjusts the operation plan of different response devices inside accordingly, ultimately enabling smart buildings to participate in demand-side response with the load baseline as the target.

[0061] This application verifies the effectiveness of the proposed intelligent building load guideline generation method through simulation examples. All optimization problems are solved using the commercial Gurobi solver. This simulation uses peak shaving and valley filling as the response scenario, with load data sourced from the publicly available DataPortals dataset (a multi-type load dataset from the UC Irvine School of Information and Computer Science). The control characteristic parameters of each intelligent building cluster are randomly sampled from a uniform distribution with set upper and lower limits (using a fixed random number seed to ensure reproducibility). Simulation time intervals are specified. The total cycle is 15 minutes. The time interval is 24 hours; since the load in the original dataset is sampled every 1 hour, interpolation is performed after clustering the load curve characteristics to ensure the consistency of the time interval. The example includes 6 smart building clusters, each containing 10 smart buildings. The advantages of the proposed method are verified from three aspects: the aggregation effect of cluster response capability, the reachability of cluster response, and the decomposition effect of cluster guidelines.

[0062] This application embeds the ideal load profile of a smart building cluster into the cluster response capability aggregation process, enabling comprehensive cluster response capability aggregation. First, the historical load data of each smart building in the cluster is cleaned and invalid data is removed. Then, the load curves of the smart buildings are normalized and load features (climbing features and edge point features) are extracted. Clustering is then performed based on these load features, and the load curve clustering results are as follows: Figure 3 As shown, typical curves for each type of load are also generated, such as... Figure 4 As shown; setting the cluster number to 1 allows for the clustering of the entire intelligent building cluster, yielding a typical load curve for the intelligent building cluster, as shown in the figure. Figure 5 As shown.

[0063] After completing the analysis of the load curve of the intelligent building, it is necessary to generate the corresponding ideal load guideline according to different scenarios. Taking the peak shaving and valley filling scenario as an example, the ideal load guideline is obtained by solving formula (10), and the ideal load guideline of the intelligent building cluster is obtained as follows. Figure 6 As shown; then, using this ideal guideline as the target, the response capabilities of the intelligent building cluster are aggregated, and the aggregation result of the intelligent building cluster response is as follows. Figure 7 As shown.

[0064] Compared to traditional aggregation methods that "maximize external response" (such as...) Figure 8 As shown in the figure, although the traditional method has a larger overall response volume, the peak response capability decreases by 78.45% in specific scenarios. The root cause is that the response subjects within the cluster have time coupling characteristics (some subjects cannot continue to participate after a limited number of responses); the traditional method aggregates targets covering the entire time period, while this application achieves targeted aggregation by optimizing the response timing, making full use of the response resources within the cluster, and achieving better aggregation results in specific scenarios.

[0065] Based on cluster response capabilities, a robust optimization model considering response randomness is constructed to generate response curves for intelligent building clusters. The response curves for intelligent building clusters are shown below. Figure 9 As shown. Taking response cluster 1 as an example, the cluster load guideline of the intelligent building cluster is generated by combining the typical load characteristic curve of the intelligent building cluster, and the load guideline of the intelligent building cluster under the peak shaving and valley filling scenario is obtained as follows. Figure 10 As shown; then, the cluster load profile is decomposed according to the load curve type to obtain the decomposition result of the intelligent building cluster load profile as shown. Figure 11 As shown. Taking cluster 1 as an example, the cluster load profile is decomposed to obtain various intelligent building load profile diagrams, as shown below. Figure 12 As shown, the parts of each type of smart building that need to be responded to are as follows: Figure 12 The blue area is shown in the image.

[0066] To verify the accessibility improvement effect, the cluster load baseline without considering the intelligent building control characteristics is compared with the cluster load baseline considering response capability in this application. The resulting intelligent building cluster load baseline accessibility comparison diagram is shown below. Figure 13 As shown; load guidelines for various intelligent buildings without considering regulation characteristics, such as Figure 14 As shown; because the intelligent building response curve obtained by decomposing the cluster load guideline without considering the control characteristics of intelligent buildings does not consider the control characteristics, the guideline is generated entirely according to the system requirements, and some areas exceed the maximum response capacity; while the guideline of the method in this application is within the response capacity range, which can effectively realize the response and significantly improve accessibility. It is worth noting that because the load guideline is a normalized curve, the decomposition process is difficult to directly incorporate the response capacity of each type of building, which will dilute accessibility to a certain extent (essentially because the guideline, as a shape curve, cannot be directly associated with the actual response amount); if the normalized guideline in formula (20) is restored to the actual value and the response capacity constraint is added to the decomposition model, although the accessibility dilution can be eliminated, the effect of reducing adjustment costs will be reduced. Therefore, in practical applications, it is necessary to balance the load adjustment cost and the accessibility of the guideline.

[0067] The results of comparing the response values ​​(which reflect control costs) of directly using the cluster load baseline and the decomposed intelligent building load baseline are shown in Table 1. Table 1 Comparison of Load Guidelines Before and After Decomposition in Smart Building Clusters

[0068] As shown in Table 1, the adjustment amount decreased significantly after decomposition of various building types, with the total adjustment amount decreasing by 92.00% (e.g., the adjustment amount of the first type of building decreased from 0.1371 to 0.0103, a decrease of 92.45%). Verification showed that the response amounts of all building types were within the response capacity range. The significant reduction in adjustment costs is due to the fact that the second type of load accounted for a large proportion in the example, and the cluster guideline is similar to the characteristics of this type of load, leading to higher adjustment amounts for other building types. Guideline decomposition allows for targeted allocation of response tasks, reducing the overall adjustment burden. If the proportion of each building type is adjusted to a uniform distribution, the total adjustment amount can still be reduced by 66.13%, demonstrating a significant cost reduction effect.

[0069] To address the problems in related technologies, such as the varying load curves of different smart buildings making it difficult to fully reflect the characteristics of the overall cluster and its various internal load curve types, the lack of consideration for the load regulation characteristics in load guideline formulation, and the failure to effectively decompose load guidelines among different types of smart buildings within the cluster, this application proposes the following solutions: First, a method for extracting and clustering the features of smart building load curves is proposed. By extracting features such as inflection points and ramp points of load curves and performing clustering, typical load curves of smart building clusters and load shape curves of various types of smart buildings are generated, fully describing the load curve characteristics of the cluster and its various internal buildings. Second, a method for generating smart building cluster load guidelines that considers regulation characteristics is proposed. First, the response curve generated by the response demand is superimposed with the typical load curve to obtain an ideal load guideline. Then, the response capability of the smart building cluster is aggregated in a targeted manner with the ideal load guideline as the target. The aggregated response capability curve is embedded into the actual load guideline formulation process to fully explore the response capability of the cluster under different response scenarios. Third, a cluster load guideline decomposition method with the goal of minimizing the total amount of smart building regulation is proposed. By reasonably decomposing the smart building cluster load guideline, the problem of decomposing load guidelines of various types of smart buildings is effectively solved, reducing the response cost of each smart building.

[0070] Please see Figure 15 This application also provides an intelligent building load guideline generation device, which can implement the above-mentioned intelligent building load guideline generation method. The device includes: The extraction module 1501 is used to extract load curve features from the historical load data of each smart building in the smart building cluster, and obtain feature vectors including ramp features and edge point features. Clustering module 1502 is used to cluster the load curves of smart buildings based on feature vectors and using clustering algorithms to generate typical load characteristic curves of smart building clusters and load shape curves of various types of smart buildings. The generation module 1503 is used to generate the ideal load guideline for the intelligent building cluster based on the system response demand curve and the typical load characteristic curve of the intelligent building cluster. The aggregation module 1504 is used to aggregate the response capabilities of intelligent building clusters by optimizing the model with the ideal load guideline as the target, and obtain the aggregated results of cluster response capabilities. Processing module 1505 is used to generate a smart building cluster load baseline based on the aggregation results of cluster response capabilities; The decomposition module 1506 is used to decompose the load profile of the intelligent building cluster into load profiles of various types of intelligent buildings with the goal of minimizing the total adjustment of each intelligent building.

[0071] The specific implementation method of the intelligent building load guideline generation device is basically the same as the specific implementation method of the intelligent building load guideline generation method described above, and will not be repeated here.

[0072] Thirdly, embodiments of this application provide an electronic device, see [link to relevant documentation]. Figure 16 The diagram shown is a structural schematic of an electronic device provided in this application. Figure 16 As shown, the device includes: a memory 31 for storing a computer program; and a processor 32 for executing the computer program; wherein, when the processor 32 executes the computer program, it implements the intelligent building load guideline generation method as described in any of the above embodiments.

[0073] For example, a computer program may be divided into one or more modules / units, one or more of which are stored in memory 31 and executed by processor 32 to complete this application. One or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in an electronic device.

[0074] The processor 32 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0075] The memory 31 can be used to store computer programs and / or modules. The processor 32 implements various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory 31 and calling the data stored in the memory 31. The memory 31 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.).

[0076] It should be noted that the aforementioned electronic devices include, but are not limited to, processors and memory, as will be understood by those skilled in the art. Figure 16 The structural diagram is merely an example of the electronic device described above and does not constitute a limitation on the electronic device. It may include more components than shown in the diagram, or combine certain components, or use different components.

[0077] Fourthly, embodiments of this application also provide a computer-readable storage medium storing a computer program. When the computer program is executed, it implements the intelligent building load guideline generation method of any of the above embodiments. It should be understood that all or part of the processes in the above-described intelligent building load guideline generation method can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the above-described intelligent building load guideline generation method.

[0078] Fifthly, embodiments of this application also provide a computer program product. The computer program product is stored in a storage medium and executed by at least one processor to implement the intelligent building load baseline generation method of any of the above embodiments. Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0079] The above description is the preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications are also considered to be within the scope of protection of this application.

Claims

1. A method for generating load guidelines in intelligent buildings, characterized in that, include: The historical load data of each smart building in the smart building cluster are used to extract load curve features, resulting in feature vectors including ramp features and edge point features. Based on the feature vector, clustering algorithms are used to cluster the load curves of smart buildings to generate typical load characteristic curves of smart building clusters and load shape curves of various types of smart buildings. Based on the system response demand curve and the typical load characteristic curve of the intelligent building cluster, an ideal load guideline for the intelligent building cluster is generated. Using the ideal load guideline as the target, the cluster response capabilities of intelligent buildings are aggregated through optimization models to obtain the cluster response capability aggregation results; Based on the aggregated results of the cluster response capabilities, a load baseline for the intelligent building cluster is generated. With the goal of minimizing the total adjustment of each smart building, the load baseline of the smart building cluster is decomposed into load baselines for each type of smart building.

2. The intelligent building load guideline generation method as described in claim 1, characterized in that, The load curve feature extraction includes: The load curve of the intelligent building is normalized to obtain the shape of the load curve. Extract ramp features, defined as events in which the degree of change in the load curve exceeds a first preset threshold within a set time window; Extract edge point features, defined as an event where the slope of the load curve changes beyond a second preset threshold at a certain moment; Extract routine load characteristics, including daily maximum load, daily peak-to-valley difference, daily peak-to-valley difference rate, and daily load factor.

3. The intelligent building load guideline generation method as described in claim 1, characterized in that, The clustering algorithm uses the k-means++ algorithm and determines the number of clusters using the silhouette coefficient method; the typical load characteristic curve of the intelligent building cluster is obtained by setting the number of clusters to 1.

4. The intelligent building load guideline generation method as described in claim 1, characterized in that, The process of generating the ideal load baseline for the intelligent building cluster is specifically achieved by superimposing the system response demand curve with the typical load characteristic curve of the intelligent building cluster.

5. The intelligent building load guideline generation method as described in claim 4, characterized in that, The optimization model for the response capability of the aggregated intelligent building cluster includes: Objective function: Minimize the deviation between the ideal load baseline and the actual total cluster load; Constraints include response capacity constraints, maximum response duration constraints, response frequency constraints, maximum number of responses constraints, and response speed constraints.

6. The method for generating intelligent building load guidelines as described in claim 1, characterized in that, The specific method for generating the intelligent building cluster load baseline is as follows: based on the cluster response capability aggregation results, a response curve is generated through a robust optimization model and superimposed with the typical load characteristic curve of the intelligent building cluster.

7. A smart building load guideline generation device, characterized in that, include: The extraction module is used to extract load curve features from the historical load data of each smart building in the smart building cluster, and obtain feature vectors including ramp features and edge point features. The clustering module is used to cluster the load curves of smart buildings based on the feature vectors using a clustering algorithm, and generate typical load characteristic curves of smart building clusters and load shape curves of various types of smart buildings. The generation module is used to generate an ideal load guideline for the intelligent building cluster based on the system response demand curve and the typical load characteristic curve of the intelligent building cluster. The aggregation module is used to aggregate the response capabilities of intelligent building clusters by optimizing the model, with the ideal load guideline as the target, to obtain the aggregated cluster response capabilities result; The processing module is used to generate a load baseline for the intelligent building cluster based on the aggregation results of the cluster response capabilities. The decomposition module is used to decompose the load baseline of the intelligent building cluster into load baselines of various types of intelligent buildings with the goal of minimizing the total adjustment of each intelligent building.

8. An electronic device, characterized in that, The system includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor, when executing the computer program, implements the intelligent building load guideline generation method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform the intelligent building load guideline generation method as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, The computer program product includes a computer program or computer instructions, which, when executed by a processor, implement the intelligent building load guideline generation method as described in any one of claims 1 to 6.

Images