An optimized control method for a water-cooled refrigeration station system in public buildings

Abstract

The present invention discloses an optimized control method for a water-cooled refrigeration station system for public buildings, which relates to the technical field of exhibition hall refrigeration control. The steps of the method include: performing three-dimensional dynamic grid modeling on the entire exhibition hall space, real-time monitoring the basic sensible heat and solar radiation in the exhibition hall to obtain the heat load weight, and dividing the exhibition hall into zones; setting a physical field coupling heat load prediction model according to the heat source distribution value of each exhibition hall zone, and combining air flow, heat conduction and solar radiation heat absorption of the glass curtain wall to predict the heat load prediction value of the exhibition hall zone; when the heat load threshold is reached, determining whether to proceed to the next step; installing a swirl air outlet with an angle adjustment mechanism beside the main engine room, and adjusting the air supply angle and air speed according to the heat load prediction value of the exhibition hall zone. The present invention solves the problems of instantaneous large load increase caused by direct sunlight on large-area floor-to-ceiling windows or glass curtain walls in summer, and dead air supply areas and uneven stratification caused by high-density display cabinets and temporary partitions.

Description

Technical Field

[0001] The present invention relates to the technical field of exhibition hall refrigeration control, and in particular to a water-cooling refrigeration station system optimization control method for public buildings. Background Art

[0002] Public buildings—especially large convention centers and exhibition halls—place stringent demands on cooling systems for energy efficiency and environmental comfort. Water-cooled systems often have large capacities and experience volatile loads. Inadequate control strategies can lead to high energy consumption, uneven local temperature distribution, and frequent equipment starts and stops.

[0003] Most existing projects still use an empirical static reset strategy, simply adjusting the chilled water supply temperature up and down according to the outdoor air or return water temperature; at the same time, the system generally relies on classic PID or rule control, which responds slowly to drastic fluctuations in cold and hot loads, making it difficult to achieve optimal operation of the entire plant, and often encountering operational inefficiencies such as "cold and hot stratification" and "large water volume and small temperature difference".

[0004] With the dramatic changes in exhibition scale and visitor numbers, indoor cooling loads can fluctuate several times in a short period of time, making it difficult for traditional control systems to respond to these demands. Furthermore, large floor-to-ceiling windows or glass curtain walls exposed to direct sunlight in summer can lead to a significant instantaneous increase in load. Without rapid linkage between sunshade and cooling, localized high-temperature zones can easily occur. Furthermore, complex airflow organization, high-density display cabinets, and temporary partitions frequently alter indoor airflow channels, resulting in dead corners and uneven stratification.

[0005] To this end, the present invention provides a water-cooling refrigeration station system optimization control method for public buildings. Summary of the Invention

[0006] The object of the present invention is to provide a method for optimizing and controlling a water-cooling refrigeration station system for a public building, so as to solve the existing problems raised in the above-mentioned background technology.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for optimizing and controlling a water-cooling refrigeration station system for a public building, comprising the following steps:

[0008] S1. Conduct three-dimensional dynamic grid modeling of the entire exhibition hall space, monitor the base sensible heat and solar radiation in the exhibition hall in real time, obtain heat load weights, and divide the exhibition hall into zones;

[0009] S2. Based on the heat source distribution value of each exhibition hall partition, a physical field coupled heat load prediction model is set up to combine airflow, heat conduction and solar radiation heat absorption of glass curtain walls to predict the heat load value of each exhibition hall partition;

[0010] S3. When the predicted heat load value is greater than the set heat load threshold, after running the dynamic sunshade interlock strategy, return to S2; when the predicted heat load value is less than or equal to the set heat load threshold, directly run S4;

[0011] S4. Install a swirl air outlet with an angle adjustment mechanism beside the main computer room, and adjust the air supply angle and air speed according to the predicted heat load value of the exhibition hall partition.

[0012] A further improvement of the present invention lies in that the specific steps of S1 include:

[0013] S11. Discretize the exhibition hall space into voxels according to N×M×H to obtain an exhibition hall voxel set ; Each voxel has a plane center coordinate ;

[0014] S12. Collect the basic sensible heat Qther, transient temperature rise , solar radiation load Isol and space occlusion rate Cpar of each voxel unit;

[0015] S13. After normalizing the data in step S12, perform weighted summation to obtain the heat load weight of each voxel unit ;

[0016] S14. Through load homogeneous density peak clustering, automatically divide into exhibition hall partitions according to the heat load weight.

[0017] A further improvement of the present invention lies in that the basic sensible heat is obtained by deploying a pyroelectric infrared or UWB positioning personnel counter, combining with the input of equipment power, and calculating the sensible heat power in each voxel unit in real time to obtain the basic sensible heat Qther;

[0018] Install a spectral sensor above the voxel near the window to obtain the total radiation of 400–1100nm in real time as the real-time solar irradiance G, then the solar radiation load , where represents the daylighting roof area corresponding to the voxel unit, represents the glass heat absorption coefficient.

[0019] A further improvement of the present invention lies in that the specific implementation steps of the load homogeneous density peak clustering include:

[0020] S141. Introduce a heat load difference weight and Euclidean distance to calculate the mixed distance between each voxel , where λ represents a control coefficient for controlling the relative weight of space and load;

[0021] S142. Calculate the local density of voxel i , where represents the density cutoff distance, taking all 2% quantile;

[0022] S143. Traverse the local densities of all voxels around voxel i in ascending order of distance. When the local density of a new voxel is greater than the local density of voxel i, stop traversing and extract the voxel as the candidate voxel cluster center, and extract its mixed distance with voxel i. ;

[0023] S144. Calculate the local density standard value of all voxels and hybrid distance standard value , extract satisfaction and The candidate voxel cluster center is used as the voxel cluster center voxel;

[0024] S145. For non-central voxels, add the voxel clusters of the nearest voxels with higher density to obtain the preliminary pavilion partition.

[0025] A further improvement of the present invention is that the specific implementation step of the load homogeneous density peak clustering further includes calculating the area of each voxel cluster ,like If the cluster area is smaller than the set minimum, the clusters with The spatially adjacent clusters with the smallest load mean difference are merged until the area of all clusters is larger than the set minimum cluster area, and the exhibition hall partition is obtained.

[0026] The present invention is further improved in that the specific step S2 includes:

[0027] S21. Establish a multi-physics coupling model, introduce airflow, heat conduction and glass curtain wall solar radiation absorption to couple and calculate the heat source distribution value of each exhibition hall partition;

[0028] S22. Create a physical field coupled heat load prediction model, which uses edge computing nodes to aggregate the temperature, wind speed, direct and scattered solar radiation intensity measured by spectral sensors provided by the sensor network in real time, as well as the geometry and point cloud data after BIM partitioning, and performs deep fusion and processing through the spatiotemporal graph convolutional network. The model input tensor includes five dimensions: basic sensible heat, solar radiation load, transient temperature rise, airflow velocity, and space occlusion rate, and outputs the heat load forecast value of each partition in the next 15 minutes. .

[0029] The present invention is further improved in that the dynamic sunshade interlocking strategy includes the use of continuously adjustable electrochromic glass, which is adjusted according to the degree of regional cooling overload. Calculate the transmittance drop S, represents the heat load threshold, where , Indicates the scale quantity from the heat load threshold to the maximum heat load, and calculates the target optical transmittance of any area where the predicted heat load value is greater than the set heat load threshold , Indicates the optical transmittance when fully transparent, Indicates the optical transmittance when darkest, and sends the target optical transmittance to the electrochromic glass controller.

[0030] A further improvement of the present invention lies in that, after the transmittance of the electrochromic glass changes, in step S21, the solar radiation load in the solar radiation heat absorption item of the glass curtain wall is updated to , and the new is written back to S2 for re-prediction ; if the re-predicted , then the sunshade interlock is released and the process jumps to S4.

[0031] A further improvement of the present invention lies in that S4 includes the following specific steps:

[0032] S41. Define the control objectives, including the water temperature wte, the angle of each air outlet

[0033] and the wind speed wv; S42. Randomly generate a set of PID control parameters

[0034] and the corresponding control objective output combinations;

[0035] S43. Calculate the exhibition hall temperature uniformity index Ute and the regional temperature error index E; S44. Take the PID control parameters

[0036] and the temperature uniformity index Ute and the regional temperature error index E as constraints, and define the hard constraint function Mte; S45. Extract the control objective output combination sequence corresponding to the PID control parameters that satisfy the hard constraint function Mte

[0037] ; S46. Set the soft constraint penalty function Jte, and extract the sequence .

[0038] A further improvement of the present invention lies in that the exhibition hall temperature uniformity index Ute and the regional temperature error index E are obtained by extracting the temperature of the i-th exhibition hall partition after sunshading and the set target temperature Calculated; First, calculate the normalized standard deviation of all exhibition hall partitions , and obtain the exhibition hall temperature uniformity index ; The regional temperature error index E is obtained by measuring the deviation of the temperature from the target temperature , then .

[0039] The further improvement of the present invention lies in that the constraint formula of the hard constraint function Mte is , where s.t. means satisfying simultaneously; the constraint formula of the soft constraint penalty function Jte is:

[0040] ;

[0041] Among them, represents the constraint weight of the temperature error term, represents the constraint weight of the temperature uniformity term.

[0042] Compared with the prior art, the beneficial effects of the present invention are:

[0043] Firstly, the present invention automatically divides the exhibition hall partitions with homogeneous load through the load homogeneous density peak clustering algorithm, combining "hybrid distance" and "minimum area constraint", solving the problem that the traditional air-conditioning zoning of exhibition halls is rough and static, and cannot cope with the uneven heat load caused by the replacement of exhibits, the change of crowd density and the change of light;

[0044] Secondly, a multi-physical field heat load prediction model coupling air flow - heat conduction - solar radiation heat absorption is constructed, and combined with a spatio-temporal graph convolutional network, the heat load prediction for the next 15 minutes is realized, with high prediction accuracy and timely response, and the electrochromic glass is used to dynamically link and change the light transmittance according to the heat load, shading by region grading, avoiding cold load overload and improving energy efficiency; after the shading adjustment, the heat load prediction model is fed back in real time, forming a closed-loop control link of prediction - linkage - feedback - correction; solving the problem that the shading strategy is out of touch with the actual load, and it is easy to have the problem of over-shading or insufficient shading;

[0045] Finally, based on the PID parameter iterative search and optimization, the dual constraints of temperature uniformity and regional error index are defined to balance the overall comfort and local accuracy; solving the problem that the fixed air supply angle and wind speed cannot adapt to the dynamic heat load differences of partitions, resulting in local cold and heat unevenness. Description of the Drawings

[0046] Figure 1 is the flow chart of an optimized control method for a water-cooled refrigeration station system for public buildings according to the present invention;

[0047] Figure 2 is the flow chart of a physical field coupling heat load prediction model according to the present invention;

[0048] Figure 3 This is a flowchart of the air supply angle and air velocity adjustment method for an optimized control method of a water-cooled refrigeration station system for public buildings according to the present invention. Specific embodiments

[0049] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations on the technical solution of the present invention. Without conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0050] The term "and / or" is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent three situations: A exists alone, A and B exist simultaneously, and B exists alone.

[0051] Embodiment 1

[0052] Figure 1 A flowchart of an optimized control method for a water-cooled refrigeration station system for public buildings disclosed in this embodiment is shown as follows:

[0053] S1. Perform three-dimensional dynamic grid modeling on the entire exhibition hall space, monitor the basic sensible heat and solar radiation in the exhibition hall in real time, obtain the heat load weight, and divide the exhibition hall into zones; the specific steps include:

[0054] S11. Install Velodyne VLP-16 radars at the main roads and high altitudes in the exhibition hall, with a horizontal accuracy of ±3 cm and a scanning frequency of 20 Hz. Register the real-time point cloud with the BIM (Revit / IFC) model through the ICP algorithm, remove static obstacles such as the ground and stairs, and only retain display cabinets and temporary partitions; discretize the exhibition hall space into voxels according to N×M×H to obtain the exhibition hall voxel set ; each voxel has a plane center coordinate ;

[0055] S12. Collect the basic sensible heat Qther, transient temperature rise , solar radiation load Isol, and space occlusion rate Cpar of each voxel unit;

[0056] The basic sensible heat is obtained by deploying pyroelectric infrared or UWB positioning personnel counters, combining with the input of equipment power and reading the socket circuit of the exhibition area through Modbus, and calculating the sensible heat power in each voxel unit in real time to obtain the basic sensible heat Qther; and place a temperature sensor at the center of each voxel, sample once every 5 s, and calculate the instantaneous temperature rise rate, that is, the transient temperature rise ; Install a spectral sensor (Hamamatsu C12880MA) above the window element to obtain the total radiation in the range of 400–1100nm in real time as the real-time solar irradiance G; then the solar radiation load , where represents the daylighting roof area corresponding to the voxel unit, represents the glass heat absorption coefficient; the laser point cloud data calculates the proportion of the point cloud within the voxel cone through PCL to obtain the spatial occlusion rate Cpar;

[0057] S13. Normalize and sum the above four types of data with weights to obtain the heat load weight of each voxel unit ;

[0058] S14. Then call the DPC algorithm based on k-d tree accelerated nearest neighbor search to form a load homogeneous density peak clustering, and automatically segment it into exhibition hall partitions according to the heat load weight, ensuring that the load of each area is homogeneous and the area is not less than 5m²; finally, write the partition number into the BMS partition table to realize the association with the subsequent controller;

[0059] The specific implementation steps of the load homogeneous density peak clustering include:

[0060] S141. To balance spatial adjacency and load similarity, introduce a "hybrid distance" and calculate the hybrid distance between each voxel through the heat load difference weight and the Euclidean distance , where λ represents a control coefficient used to control the relative weight of space and load; to make their dimensions consistent;

[0061] S142. Calculate the local density of voxel i , where represents the density truncation distance, taking the 2% quantile of all ; this Gaussian kernel density can smoothly balance the influence of several surrounding neighboring points;

[0062] S143. Traverse the local density of all voxels around voxel i in ascending order of distance. When a new voxel with a local density greater than that of voxel i is traversed, stop the traversal, extract this voxel as the candidate voxel cluster center, and extract its hybrid distance from voxel i ;

[0063] S144. Calculate the standard values of the local density and the standard value of the hybrid distance of all voxels, and extract the candidate voxel cluster center that satisfies and as the voxel cluster center voxel;

[0064] S145. For non-central voxels, add the voxel clusters of the nearest voxels with higher density to obtain the preliminary pavilion partitioning;

[0065] S146. Calculate the area of each voxel cluster ,like If the cluster area is smaller than the set minimum, the clusters with The spatially adjacent clusters with the smallest load mean difference are merged until the area of all clusters is larger than the set minimum cluster area, and the exhibition hall partition is obtained.

[0066] S2. Based on the heat source distribution value of each exhibition hall partition, a physical field coupled heat load prediction model is set up to combine airflow, heat conduction and solar radiation heat absorption of glass curtain walls to predict the heat load value of each exhibition hall partition;

[0067] S3: When the predicted heat load value is greater than the set heat load threshold, the dynamic sunshade interlocking strategy is executed and the system returns to S2; when the predicted heat load value is less than or equal to the set heat load threshold, S4 is executed directly;

[0068] S4. Install a swirl air outlet with an angle adjustment mechanism next to the main machine room, and adjust the air supply angle and wind speed according to the predicted heat load value of the exhibition hall partition.

[0069] Example 2

[0070] Based on the same inventive concept as Example 1, this embodiment provides a specific implementation of a physical field coupled heat load prediction model and a dynamic sunshade interlocking strategy, which is used to implement the implementation steps of step S2 and step S3 in Example 1. Figure 2 A flow chart of a physical field coupled heat load prediction model of this embodiment is shown, and the specific steps include:

[0071] S21. Establish a multi-physics coupling model, introduce the coupling of airflow, heat conduction and solar radiation absorption of glass curtain wall to calculate the heat source distribution value of each exhibition hall partition; the specific formula is expressed as follows: , T represents the temperature of the exhibition hall partition, where, Describe the effects of heat conduction, represents the target temperature, h represents the convective heat transfer coefficient, and the empirical value is removed. ;

[0072] S22. Create a physical field coupled heat load prediction model, which uses edge computing nodes to aggregate the temperature, wind speed, direct and scattered solar radiation intensity measured by spectral sensors provided by the sensor network in real time, as well as the geometry and point cloud data after BIM partitioning, and performs deep fusion and processing through the spatiotemporal graph convolutional network. The model input tensor includes five dimensions: basic sensible heat, solar radiation load, transient temperature rise, airflow velocity, and space occlusion rate, and outputs the heat load forecast value of each partition in the next 15 minutes. 。

[0073] The dynamic sunshade interlock strategy includes using continuously adjustable electrochromic glass, whose optical transmittance ranges from fully transparent to the darkest and can vary linearly. According to the degree of regional cooling overload calculate the transmittance reduction S, which represents the heat load threshold, where , represents the scale from the heat load threshold to the maximum heat load, and calculate the target optical transmittance of any area where the predicted heat load value is greater than the set heat load threshold , represents the optical transmittance at full transparency, represents the optical transmittance at the darkest state. Send the target optical transmittance to the electrochromic glass controller. After the transmittance of the electrochromic glass changes, the solar radiation load in the solar radiation heat absorption term of the glass curtain wall in step S21 is updated to , and write the new back to S2, and re-predict ; if after re-prediction, then the sunshade interlock is released, and jump to S4.

[0074] Embodiment 3

[0075] Figure 3 shows a flow chart of the air supply angle and air speed adjustment method for an optimized control method of a water-cooled refrigeration station system for public buildings in this embodiment. Based on the inventive concepts of Embodiment 1 and Embodiment 2, this embodiment provides an air supply angle and air speed adjustment method for realizing the implementation steps of step S4 in Embodiment 1. The specific steps include:

[0076] S41. Define the control objectives, including the water temperature wte, the angle of each air outlet

[0077] and the air speed wv; and the corresponding control objective output combination;

[0078] S43. Calculate the exhibition hall temperature uniformity index Ute and the regional temperature error index E;

[0079] The exhibition hall temperature uniformity index Ute and the regional temperature error index E are calculated by extracting the temperature of the i-th exhibition hall partition after sunshade and the set target temperature , the maximum temperature difference between partitions ; First calculate the normalized standard deviation of all exhibition hall partitions , the uniformity index of the exhibition hall temperature is obtained , so that , where , represents the average temperature of all exhibition hall partitions, and the requirement is , represents the minimum value of the required temperature uniformity; the regional temperature error index E is obtained by measuring the deviation of the temperature from the target temperature , then .

[0080] S44. Take the PID control parameters and the temperature uniformity index Ute and the regional temperature error index E as constraints, and define the hard constraint function Mte; make the temperature uniformity of the exhibition hall greater than the lowest uniformity threshold, and the temperature close to the target temperature; the constraint formula of the hard constraint function Mte is , where s.t. means satisfying simultaneously;

[0081] S45. Extract the control target output combination sequence corresponding to the PID control parameters that satisfy the hard constraint function Mte ;

[0082] S46. Set the soft constraint penalty function Jte:

[0083] ;

[0084] where represents the constraint weight of the temperature error term, represents the constraint weight of the temperature uniformity term.

[0085] Extract the sequence The control target output combination sequence corresponding to the minimum value of the soft constraint penalty function Jte in is used as the final output .

[0086] By simultaneously monitoring the "overall temperature difference distribution" (through ) and the "average deviation" (through E), avoid simply pursuing the average value and covering up local hot spots / cold spots.

[0087] The set values such as the threshold and weight can be set according to the default settings of the present invention, or can be set by the operator himself.

[0088] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) that contain computer-usable program code.

[0089] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0090] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing devices to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that realizes the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0091] These computer program instructions can also be loaded onto a computer or other programmable data processing devices, such that a series of operation steps are executed on the computer or other programmable devices to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable devices provide steps for realizing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0092] The embodiments of the present invention have been described above in conjunction with the accompanying drawings. However, the present invention is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present invention, those of ordinary skill in the art can also make many forms without departing from the spirit and scope protected by the present invention and the claims. These all fall within the protection scope of the present invention.

Claims

1. An optimized control method for a water-cooled refrigeration station system used in public buildings, characterized in that: It includes the following steps: S1. Conduct three-dimensional dynamic grid modeling on the entire exhibition hall space, monitor the basic sensible heat and solar radiation in the exhibition hall in real time, obtain the heat load weight, and divide the exhibition hall into zones; S2. Set up a physical field coupling heat load prediction model according to the heat source distribution value of each exhibition hall zone, and combine air flow, heat conduction and solar radiation heat absorption of the glass curtain wall to predict the heat load prediction value of the exhibition hall zone; S3. When the heat load prediction value is greater than the set heat load threshold, run the dynamic sunshade interlock strategy and then return to S2; when the heat load prediction value is less than or equal to the set heat load threshold, directly run S4; S4. Install a swirl air outlet with an angle adjustment mechanism beside the main computer room, and adjust the air supply angle and air speed according to the heat load prediction value of the exhibition hall zone; The specific steps of S1 include: S11. Discretize the exhibition hall space into voxels of N×M×H to obtain the exhibition hall voxel set ; Each voxel has a planar center coordinate ; S12. Collect the basic sensible heat Qther and transient temperature rise of each voxel unit , solar radiation load Isol, and spatial occlusion rate Cpar; S13. Normalize the data in step S12 and then perform weighted summation to obtain the heat load weight of each voxel unit ; S14. Automatically divide into exhibition hall zones according to the heat load weight through load homogeneous density peak clustering; The specific implementation steps of the load homogeneous density peak clustering include: S141. Introduce the heat load difference weight and Euclidean distance to calculate the mixed distance between each voxel , where λ represents the control coefficient, which is used to control the coefficient of the relative weight of space and load; S142. Calculate the local density of voxel i , where represents the density truncation distance, taking the 2% quantile of all ; S143. Traverse the local densities of all voxels around voxel i in ascending order of distance. When a new voxel with a local density greater than that of voxel i is traversed, stop the traversal, extract this voxel as the candidate voxel cluster center, and extract the mixing distance between it and voxel i ; S144. Calculate the local density standard values of all voxels and the standard values of the mixing distance , and extract the candidate voxel cluster centers that satisfy and as the voxel cluster center voxels; S145. For non-central voxels, add them to the voxel cluster of the voxel that is the nearest and has a higher density to obtain a preliminary exhibition hall zone.

2. The optimized control method for a water-cooled refrigeration station system for public buildings according to claim 1, characterized in that: The basic sensible heat is obtained by deploying a pyroelectric infrared or UWB positioning personnel counter, combining with the input of equipment power, and calculating the sensible heat power in each voxel unit in real time to obtain the basic sensible heat Qther; Install a spectral sensor above the window element to obtain the total radiation in the range of 400–1100 nm in real time as the real-time solar irradiance G, then the solar radiation load , where represents the daylighting roof area corresponding to the voxel unit, represents the glass heat absorption coefficient.

3. A method for optimizing the control of a water-cooled refrigeration station system for public buildings according to claim 1, characterized in that: The load homogeneity density peak clustering step further includes calculating the area of each voxel cluster ,like If the cluster area is smaller than the set minimum, the clusters with The spatially adjacent clusters with the smallest load mean difference are merged until the area of all clusters is larger than the set minimum cluster area, and the exhibition hall partition is obtained.

4. The optimized control method for a water-cooled refrigeration station system for public buildings according to claim 1, characterized in that: The specific steps of S2 include: S21. Establish a multi-physical field coupling model, introduce the coupling of air flow, heat conduction and solar radiation heat absorption of the glass curtain wall to calculate the heat source distribution value of each exhibition hall zone; S22. Create a physical field coupled heat load prediction model. Real-time summarize the temperature, wind speed, direct and diffuse solar radiation intensity measured by spectral sensors, and the geometric and point cloud data after BIM zoning provided by the sensor network through the edge computing node, and perform deep fusion and processing through the spatio-temporal graph convolutional network. The model input tensor includes five dimensions: basic sensible heat, solar radiation load, transient temperature rise, air flow velocity, and spatial occlusion rate, and outputs the heat load prediction values of each zone within the next 15 minutes. .

5. The optimized control method for a water-cooled refrigeration station system for public buildings according to claim 1, characterized in that: The dynamic shading interlock strategy includes using continuously adjustable electrochromic glass, and according to the degree of regional cooling overload calculate the reduction S of the transmittance, denotes the heat load threshold, where , denotes the scale amount from the heat load threshold to the maximum heat load, and calculate the target optical transmittance of any area where the predicted heat load value is greater than the set heat load threshold , denotes the optical transmittance at full transparency, denotes the optical transmittance at the darkest state, and send the target optical transmittance to the electrochromic glass controller.

6. The optimized control method for a water-cooled refrigeration station system for public buildings according to claim 5, characterized in that: The dynamic shading interlock strategy further includes that after the transmittance of the electrochromic glass changes, the solar radiation load in the solar radiation heat absorption term of the glass curtain wall in step S21 is updated to , and the new is written back to S2, and re-prediction is performed ; if the result after re-prediction , the shading interlock is released, and the process jumps to S4.

7. A method for optimizing the control of a water-cooled refrigeration station system for public buildings according to claim 1, characterized in that: S4 includes the following specific steps: S41. Define control objectives, including water temperature wte, the angle of each air outlet and wind speed wv; S42. Randomly generate a set of PID control parameters and the corresponding control target output combination; S43. Calculate the exhibition hall temperature uniformity index Ute and the regional temperature error index E; S44. Take the PID control parameters and the temperature uniformity index Ute and the regional temperature error index E as constraints, and define the hard constraint function Mte; S45. Extract the control target output combination sequence corresponding to the PID control parameters that satisfy the hard constraint function Mte ; S46. Set the soft constraint penalty function Jte and extract the sequence The control target output combination sequence corresponding to the minimum value of the soft constraint penalty function Jte in is used as the final output .

8. A method for optimizing the control of a water-cooled refrigeration station system for public buildings according to claim 7, characterized in that: The exhibition hall temperature uniformity index Ute and the regional temperature error index E are obtained by extracting the temperature of the i-th exhibition hall partition after shading and the set target temperature , the maximum temperature difference between partitions is calculated; first, calculate the normalized standard deviation of all exhibition hall partitions , to obtain the exhibition hall temperature uniformity index ; the regional temperature error index E is obtained by measuring the deviation of the temperature from the target temperature , then .

9. The optimized control method for a water-cooled refrigeration station system for public buildings according to claim 8, characterized in that: The constraint formula of the hard constraint function Mte is , where s.t. means simultaneously satisfy; the constraint formula of the soft constraint penalty function Jte is: ； Among them, represents the constraint weight of the temperature error term, represents the constraint weight of the temperature uniformity term.

Images