A clustering-based automatic division method for hot zones of an air conditioning system

Abstract

The application discloses a clustering-based automatic heat zone division method for an air conditioning system, which is characterized by the following steps: clustering analysis is performed according to the cooling and heating loads of rooms in a building; and the rooms in the building are divided into heat zones by analyzing the clustering results, the actual room layout and the type of the air conditioning system, that is, the rooms with the same load characteristics are divided into the same heat zone. The application provides an automatic division method, combines the cooling and heating load characteristics of the rooms in the building with the building layout information, and makes the heat zone division scheme of the air conditioning system meet the actual building demand, so that the air conditioning system is energy-saving, and the overcooling or overheating situation is avoided. Compared with the current zoning method in the design of the air conditioning system, the application has the advantages that the zoning process is automated and data-based, the heat zone division is more effective and reasonable by combining the load characteristics of the rooms, the geometric information and the type of the air conditioning system, and the unreasonable design of the air conditioning system caused by subjective factors of designers is avoided.

Description

Technical Field

[0001] This invention belongs to the field of HVAC design, specifically relating to the field of thermal zone division in air conditioning systems. Background Technology

[0002] In the design of air conditioning systems, the division of thermal zones is a crucial step. A thermal zone refers to a collection of one or more spaces with similar heat requirements (the same temperature control point). From the perspective of building energy consumption models, a thermal zone is the most basic thermal unit in the physical model of energy consumption simulation. From a system design perspective, a thermal zone defines the spatial areas within a building that require independent control. Reasonable thermal zone division, which groups areas with similar functional requirements and thermal characteristics into the same thermal zone, helps in the design and operation control of the air conditioning system, enabling it to meet the comfort requirements of each room while reducing unnecessary energy consumption. In traditional air conditioning system design, designers often divide thermal zones based on the functions, settings, load characteristics, and geometric information of the rooms within the building, combined with their own experience. This method is highly subjective, and the design may not necessarily conform to the actual building conditions. Existing research on automatic thermal zone division algorithms is mostly based on the simplification of building energy consumption models, rather than on the zoning of air conditioning systems. Currently, with the development of BIM and the widespread adoption of various machine learning algorithms, the thermal zone division of air conditioning systems can be automated. The building information stored in BIM can provide room geometry information and fire compartment information, while also providing a basis for load calculation; while machine learning algorithms mine the heating and cooling load characteristics of each room in the building from a data perspective, and can extract load characteristics that were not originally considered by human subjective experience, making the zoning results more scientific. Summary of the Invention

[0003] To provide a more efficient, scientific, and reasonable hot zone division scheme for air conditioning systems, enabling energy-saving operation of air conditioning systems and avoiding overcooling or overheating, this invention proposes an automatic hot zone division method for air conditioning systems based on clustering. This method combines the cooling and heating load characteristics of each room in the building with the building's floor plan information to ensure that the hot zone division scheme for the air conditioning system meets the actual building requirements.

[0004] Technical solution

[0005] A cluster-based automatic hot zone division method for air conditioning systems is characterized by performing cluster analysis based on the heating and cooling loads of each room in a building. By analyzing the clustering results, the actual room layout, and the type of air conditioning system, the building rooms are divided into hot zones; rooms with similar load characteristics are grouped into the same hot zone. Compared with existing technologies, this invention achieves automatic hot zone division based on load characteristics, enabling more efficient, reasonable, and scientific hot zone division of air conditioning systems, reducing repetitive work, and avoiding deviations caused by subjective human factors.

[0006] A clustering-based method for automatic hot zone division in an air conditioning system, characterized by the following specific implementation steps:

[0007] Step 1: Perform cluster analysis based on the heating and cooling loads of each room in the building, which will be used to inform Step 2.

[0008] Step 2: By analyzing the clustering results, the actual room layout, and the type of air conditioning system, the building rooms are automatically divided into thermal zones.

[0009] Further, in step 1, the hourly cooling and heating loads of each room in the building throughout the year are summed, with cooling load being positive and heating load being negative, and then divided by the room area to obtain the hourly load per unit area of ​​each room throughout the year. Then, by performing cluster analysis on the hourly load per unit area of ​​rooms on each floor throughout the year, rooms with similar load characteristics are grouped into the same category.

[0010] The clustering analysis method uses the k-means clustering algorithm: first, k objects are randomly selected from the data objects as initial cluster centers. For the remaining objects, they are assigned to the most similar clusters according to their Euclidean distance from these cluster centers (Equation 1). Then, the mean of all data objects in each cluster is recalculated as the new cluster center until the assignment is stable.

[0011]

[0012] In the formula, D is the sum of squared errors of all objects in the dataset; p is a point in Euclidean space (in this invention, it is the flow of people in 24 hours of a day); c i For cluster C i Centroids (p and c) i All data are 24-dimensional; the number of clusters k ranges from 2 to 20, and the value of k is determined by the Calinski-Harabasz criterion.

[0013] Furthermore, step 2:

[0014] Step 2.1: Using the "Automatic Load Hierarchy Zoning" module algorithm, the load clustering results from Step 1 are combined with the building room layout information (including room geometric coordinates, fire compartment markings, and dimensions of air ducts and machine rooms) to automatically divide hot zones. Rooms with similar load characteristics, close geometric locations, and located within the same fire compartment are grouped into the same hot zone, ensuring that the zoning results meet the actual engineering piping requirements.

[0015] The "Automatic Load Level Zoning" module algorithm performs a comprehensive analysis of room load characteristic differences and geometric proximity, that is, it optimizes the combination of load characteristic differences and geometric proximity. The combined optimization equation is shown in equation (2):

[0016]

[0017] Q = ∑ m ∑ n f(m,n) m≠n Equation (3)

[0018] In the formula, P is the objective function for optimization;

[0019] N is the total number of partitions, 1≤k≤6;

[0020] Q i The calculation formula for the load characteristics difference of the room in the i-th group is shown in Equation (3), where m and n are two rooms in the i-th group. If m and n belong to the same clustering partition in step 1, then f(m,n) = 0, otherwise f(m,n) = 1.

[0021] L i Let be the geometric proximity of the i-th group, and let be the sum of the Euclidean distances between the center points of the i-th group of rooms;

[0022] w1 and w2 are the load and geometric similarity weight coefficients, respectively.

[0023] By solving the optimization problem of equation (2), the following can be achieved:

[0024] (1) Similar load characteristics, but some rooms are geometrically close but not adjacent;

[0025] (2) Zoning issues in special cases where load changes are different but the geometric locations are adjacent and the room load is not large.

[0026] Step 2.2: Using the "System-Level Automatic Zoning" module algorithm, the load-level zoning results are adjusted according to the selected air conditioning system type to ensure the zoning results meet the actual system settings requirements. The air conditioning system types involved include VAV systems, CAV systems, fan coil units + fresh air systems, VRV systems, radiant terminal systems, and combinations of these systems.

[0027] The "system-level automatic zoning" module algorithm performs further zoning for different air conditioning systems. The processing scheme is shown in the table below:

[0028] Table 1 Zoning Scheme for Different Air Conditioning Systems

[0029]

[0030]

[0031] Compared with the current zoning methods in air conditioning system design, the advantage of this invention is that it automates and digitizes the zoning process, and combines room load characteristics, geometric information and air conditioning system type to more effectively and rationally divide the hot zones, avoiding irrational air conditioning system design caused by the designer's subjective factors. Attached Figure Description

[0032] Figure 1 This is a flowchart illustrating an automatic hot zone division method for an air conditioning system based on clustering, according to the present invention.

[0033] Figure 2 This is a building floor plan diagram in an embodiment of the present invention.

[0034] Figure 3 This is a planar schematic diagram of the load clustering results in an embodiment of the present invention.

[0035] Figure 4 The examples show the cluster centers for various summer day loads. Detailed Implementation

[0036] The embodiments of the present invention are described in detail below. These embodiments are implemented based on the technical solution of the present invention, and provide detailed implementation methods and specific operation processes. However, the scope of protection of the present invention is not limited to the following embodiments.

[0037] Figure 1 This is a flowchart illustrating an automatic hot zone partitioning method for an air conditioning system based on clustering, according to the present invention. It comprises two main parts: a load clustering algorithm and an automatic partitioning algorithm. The automatic partitioning algorithm further includes two modules: a "load-level automatic partitioning algorithm" and a "system-level automatic partitioning algorithm."

[0038] Example

[0039] Figure 2 This is a typical office building floor plan for this example, with the air conditioning system set as a VAV system.

[0040] Step 1: Put Figure 2 The calculated hourly cooling and heating loads for each room in the building throughout the year are used in the clustering method to obtain the clustering results as shown below. Figure 3 As shown. Figure 3 Areas marked with the same letter indicate that the rooms are divided into the same category. There are a total of 7 categories in the diagram: A, B, C, D, E, F, and G. Figure 4 It serves as a clustering center for various summer daily loads.

[0041] Step 2: Combine the planar information, system air conditioning, and clustering results from Step 1 to divide the thermal zones.

[0042] First, the "Load Hierarchy Automatic Partitioning Algorithm" module partitions the planar information and clustering results. For example, in cluster B, rooms B1, B2, and B3 are geometrically adjacent, and room B4 is close to other rooms in cluster B, even though they are not adjacent. Considering both similarity and geometric proximity, all rooms in cluster B are divided into the same hot zone.

[0043] Secondly, the "System-Level Automatic Partitioning Algorithm" module adjusts the partitioning results obtained by the "Load-Level Automatic Partitioning Algorithm" module according to the VAV system partitioning processing scheme to obtain the final hot zone partitioning scheme.

Claims

1. A cluster-based method for automatic thermal zone division in an air conditioning system, characterized in that, Based on the heating and cooling loads of each room in the building, cluster analysis is performed. By analyzing the clustering results, the actual room layout, and the type of air conditioning system, the building's rooms are divided into thermal zones. The specific implementation steps are as follows: Step 1: Perform cluster analysis based on the heating and cooling loads of each room in the building, which will be used to inform Step 2; Step 2: By analyzing the clustering results, actual room layout, and air conditioning system type, the building rooms are automatically divided into thermal zones; Step 1 involves summing the hourly cooling and heating loads of each room in the building throughout the year, with cooling load being positive and heating load being negative, and then dividing by the room area to obtain the hourly load per unit area of ​​each room throughout the year. Then, cluster analysis is performed on the hourly load per unit area of ​​rooms on each floor throughout the year to group rooms with similar load characteristics into the same category. The cluster analysis uses the k-means clustering algorithm. First, randomly select k objects from the data objects as initial cluster centers. For the remaining objects, assign them to the clusters most similar to them based on their Euclidean distance from these cluster centers. Then the mean of all data objects in each cluster is recalculated as the new cluster center until the assignment is stable; Equation (1) In the formula, D is the sum of squared errors of all data objects in the dataset; p is a point in Euclidean space; For clusters The centroid, p and All data are 24-dimensional; the number of clusters k ranges from 2 to 20, and the value of k is determined by the Calinski-Harabasz criterion. Step 2: Step 2.1: Using the "Automatic Load Hierarchy Zoning" module algorithm, the load clustering results from Step 1 are combined with the building room floor plan information to automatically divide the hot zones; rooms with similar load characteristics and close geometric locations that are in the same fire compartment are divided into the same hot zone, so that the zoning results meet the actual engineering pipe layout requirements. Step 2.2: Using the "System-Level Automatic Partitioning" module algorithm, the load-level partitioning results are adjusted according to the selected air conditioning system type to ensure that the partitioning results meet the actual system setting requirements. The air conditioning system types involved include VAV systems, CAV systems, fan coil units + fresh air systems, VRV systems, radiant terminal systems, and combined systems of the above. The "Automatic Load Level Zoning" module algorithm performs a comprehensive analysis of room load characteristic differences and geometric proximity, that is, it optimizes the combination of load characteristic differences and geometric proximity. The combined optimization equation is shown in equation (2): Equation (2) Equation (3) In the formula To optimize the objective function; The total number of partitions, 1 ≤ ≤6; The calculation formula for the load characteristic difference of the rooms in the i-th group is shown in equation (3), where m and n are two rooms in the i-th group. If m and n belong to the same clustering partition in step 1, then ,otherwise ; Let be the geometric proximity of the i-th group, and let be the sum of the Euclidean distances between the center points of the i-th group of rooms; , These are the load and geometric similarity weight coefficients, respectively. The "system-level automatic zoning" module algorithm performs further zoning for different air conditioning systems; the processing scheme is as follows: Zoning schemes are implemented for different types of air conditioning systems, among which; If the air conditioning system is a VAV system, then it is zoned according to load characteristics and geometric proximity; If the air conditioning system is a CAV system, then it is zoned according to load characteristics and geometric proximity; If the air conditioning system is a combination of fan coil units and a fresh air system, the fresh air system is zoned according to the characteristics of the fresh air load and geometric proximity, while the fan coil water system is divided into one zone for the entire floor. If the air conditioning system is a VRV system, then it is zoned according to load characteristics and geometric proximity; If the air conditioning system is a radiant terminal system, the fresh air system is zoned according to the fresh air load characteristics and geometric proximity, and the water system is zoned for the entire floor. If the air conditioning system is a modular air conditioning system, then each system is divided into zones according to load characteristics and geometric proximity.

Images