A data construction method, device and medium for central air conditioning unit modeling

Abstract

The embodiment of the present application discloses a data construction method, device and medium for central air-conditioning unit modeling. It belongs to the field of data processing technology and solves the problem of limited sample size during the training of central air-conditioning unit models, resulting in inaccurate model parameters. Feature points are extracted from building structure data to construct an undirected graph of building buildings, and a building model is constructed through a graph neural network and the undirected graph of building buildings; initial operating parameters are obtained, the building model is partitioned, and the obtained initial operating parameters are matched to each block; the building model is simulated and run, and the environmental changes and air-conditioning energy consumption requirements of each block in a preset time period in the future are predicted through a time series prediction model; the obtained simulation operation data and prediction data are tested and fed back to adjust the building model, so that the data output by the adjusted building model is applied to the modeling of the building central air-conditioning unit.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a data construction method, device and medium for modeling a central air-conditioning unit. Background Art

[0002] In the current field of intelligent buildings and energy management, accurate modeling of building central air-conditioning units is a key link in achieving efficient operation, energy conservation and emission reduction, and optimizing management strategies. However, in the face of increasingly complex building environments and diverse user needs, traditional modeling methods often seem to be unable to cope with the lack of data, especially insufficient historical data.

[0003] Due to the lack of sufficient historical data in the existing technology, the sample size in the central air-conditioning unit model training process is limited, and it is impossible to fully cover various operating scenarios and extreme conditions of the system. This not only increases the risk of model overfitting, but also causes inaccurate model parameter estimation when the model faces unknown or new operating conditions, thereby affecting the performance evaluation and regulation effect of the central air-conditioning unit model as a whole system. Summary of the invention

[0004] The embodiments of the present application provide a data construction method, device and medium for central air-conditioning unit modeling, which are used to solve the following technical problems: due to the lack of sufficient historical data, the sample size in the central air-conditioning unit model training process is limited, resulting in inaccurate model parameter estimation, which in turn affects the performance evaluation and control effect of the overall system of the central air-conditioning unit model.

[0005] The present application embodiment adopts the following technical solutions:

[0006] The embodiment of the present application provides a data construction method for modeling a central air-conditioning unit. The method includes: obtaining building structure data, extracting feature points from the building structure data, constructing a building undirected graph based on the extracted feature points, determining the feature point connection relationship corresponding to the building undirected graph through a graph neural network, and constructing a building model based on the building undirected graph and the feature point connection relationship; obtaining initial operation parameters, partitioning the building model, and matching the obtained initial operation parameters to each block; simulating the building model based on the initial operation parameters corresponding to the building model and each block, and predicting the environmental changes and air-conditioning energy consumption requirements of each block in a preset time period in the future through a time series prediction model corresponding to a long short-term memory network; and performing detection and feedback on the obtained simulation operation data and prediction data to adjust the building model so that the data output by the adjusted building model is applied to the modeling of the building central air-conditioning unit.

[0007] The embodiment of the present application can intuitively represent the structure and layout of a building by constructing an undirected graph of a building architecture, and more accurately determine the connection relationship between each feature point in the building through a graph neural network, providing an accurate basis for subsequent simulation operation and prediction. Through simulation operation, the operating status of the building under different conditions is simulated to provide a reference for actual operation. The long short-term memory network time series prediction model can predict environmental changes and air-conditioning energy consumption requirements of each block within a preset time period in the future, which is helpful to formulate operation strategies in advance and optimize energy use. Through detection feedback and model adjustment, the performance and accuracy of the building model are continuously optimized to make it closer to the actual operation. Based on the adjusted building model data, it is possible to obtain accurate sample data required for modeling the building central air-conditioning unit, expand the amount of modeling data, and thereby improve the accuracy of modeling the building central air-conditioning unit.

[0008] In one implementation of the present application, by acquiring building structure data, feature points of the building structure data are extracted, an undirected building graph is constructed based on the extracted feature points, and the connection relationship of the feature points corresponding to the undirected building graph is determined through a graph neural network, so as to construct a building model based on the undirected building graph and the connection relationship of the feature points. Specifically, the method includes: matching and verifying the multi-source building structure data to screen out abnormal data and obtain reference feature points; determining the correlation coefficient between the reference feature points through the Spearman rank correlation coefficient, and comparing the correlation coefficient with a preset correlation coefficient threshold to screen out reference feature point pairs that meet the correlation requirements; constructing an undirected building graph based on the reference feature point pairs; determining the connection relationship between each reference feature point pair in the undirected building graph through a graph neural network, and determining the weight between the reference feature point pairs based on the connection relationship; wherein the weight represents the close connection relationship between different nodes; and constructing a building model based on the reference feature points, the connection relationship and the weight.

[0009] In one implementation of the present application, a connection relationship between each pair of reference feature points in an undirected graph of a building is determined through a graph neural network, and the weights between the reference feature point pairs are determined based on the connection relationship, specifically including: inputting the undirected graph of the building into the graph neural network to determine the first relationship score and the connection relationship type between each pair of reference feature points through the graph neural network; and, determining the node distance corresponding to each pair of reference feature points, and determining the number of edges connected to each node; determining the second relationship score based on the node distance, and determining the third relationship score based on the number of edges; performing weighted averaging on the first relationship score, the second relationship score, the third relationship score and the score corresponding to the connection relationship type to obtain a comprehensive score; and determining the weights between the reference feature point pairs based on the comprehensive score.

[0010] In one implementation of the present application, the initial operating parameters are obtained, the building model is partitioned, and the obtained initial operating parameters are matched to each block, specifically including: obtaining a subgraph of the building undirected graph corresponding to each floor in the building model; inputting each subgraph into a preset functional feature extraction model to output a building feature graph of each floor through the preset functional feature extraction model; performing feature comparison on the building feature graph of each floor, dividing the building model into multiple blocks based on the comparison results, and determining a block position data set corresponding to the multiple blocks based on the building structure data corresponding to the subgraph; determining the initial operating parameters corresponding to the building model based on the block position data set of each floor; wherein the initial operating parameters include at least one of the initial environment data, building equipment operation data, building crowd activity data and weather data corresponding to each block; matching the initial operating parameters with the block position data set to match the initial operating parameters to each block.

[0011] In one implementation of the present application, the initial operating parameters corresponding to the building model are determined based on the block position data set of each floor, specifically including: combining the block position data set of each floor to generate a floor building structure diagram; comparing the floor building structure diagram with the reference structure diagram in the parameter database to determine a target structure diagram that meets the similarity requirements; based on the target structure diagram, determining the required parameter document in the parameter database; wherein, the parameter database is provided with multiple reference structure diagrams, and also provided with parameter documents matching the multiple reference structure diagrams.

[0012] In one implementation of the present application, after simulating the operation of the building model based on the initial operating parameters corresponding to the building model and each block, the method also includes: determining the building structure parameters and equipment placement parameters corresponding to each block, and determining the physical isolation areas corresponding to each block based on the building structure parameters and the equipment placement parameters; determining the first mutual influence parameters corresponding to different physical isolation areas within each block in a first preset influence parameter database based on the isolation object information corresponding to the physical isolation area; determining the second mutual influence parameters corresponding to each block in a second preset influence parameter database based on the distance corresponding to each block and the isolation physical information between the blocks; performing a first simulated operation adjustment on the different physical isolation areas within the block based on the first mutual influence parameters; and performing a second simulated operation adjustment on the blocks after the first adjustment based on the second mutual influence parameters.

[0013] In one implementation of the present application, the environmental changes and air-conditioning energy consumption requirements of each block within a preset time period in the future are predicted through a time series prediction model corresponding to a long short-term memory network, specifically including: obtaining simulation operation data corresponding to a building model within a preset time period; inputting the simulation operation data into a time series prediction model corresponding to a long short-term memory network to obtain equipment operation data and environmental change data corresponding to the building model within a preset time period in the future; constructing an environmental data change line graph based on the environmental change data, determining the difference between adjacent data in the environmental data change line graph, and determining an environmental change section and an environmental stability section based on the difference; determining a first air-conditioning energy consumption based on the duration of the environmental change section, and determining a second air-conditioning energy consumption based on the duration of the environmental stability section, so as to obtain a predicted value of the air-conditioning energy consumption demand based on the first air-conditioning energy consumption and the second air-conditioning energy consumption.

[0014] In one implementation of the present application, the acquired simulation operation data and prediction data are tested and fed back to adjust the building model of the building, specifically including: performing ratio processing on the duration of the environmental change section and the duration of the environmental stability section to obtain the environmental change ratio; performing ratio processing on the first air-conditioning energy consumption and the second air-conditioning energy consumption and the predicted value of the air-conditioning energy consumption demand to obtain the air-conditioning energy consumption ratio; when the environmental change ratio and the air-conditioning energy consumption ratio do not meet the preset threshold conditions, determining the first difference between the environmental change ratio and the preset threshold, and determining the second difference between the air-conditioning energy consumption ratio and the preset threshold; matching the first difference and the second difference with the model adjustment scheme database to obtain the corresponding adjustment scheme to adjust the building model.

[0015] The embodiment of the present application provides a data construction device for modeling a central air-conditioning unit, comprising: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can: obtain building structure data, extract feature points of the building structure data, construct a building undirected graph based on the extracted feature points, determine the feature point connection relationship corresponding to the building undirected graph through a graph neural network, and construct a building model based on the building undirected graph and the feature point connection relationship; obtain initial operation parameters, partition the building model, and match the obtained initial operation parameters to each block; simulate the building model based on the initial operation parameters corresponding to the building model and each block, and predict the environmental changes and air-conditioning energy consumption requirements of each block within a preset time period in the future through a time series prediction model corresponding to a long short-term memory network; detect and feedback the obtained simulated operation data and predicted data to adjust the building model, so that the data output by the adjusted building model is applied to the modeling of the building central air-conditioning unit.

[0016] A non-volatile computer storage medium provided in an embodiment of the present application stores computer executable instructions, wherein the computer executable instructions are configured to: obtain building structure data, extract feature points from the building structure data, construct an undirected graph of a building based on the extracted feature points, determine the feature point connection relationship corresponding to the undirected graph of the building through a graph neural network, and construct a building model based on the undirected graph of the building and the feature point connection relationship; obtain initial operating parameters, partition the building model, and match the obtained initial operating parameters to each block; simulate the building model based on the initial operating parameters corresponding to the building model and each block, and predict the environmental changes and air-conditioning energy consumption requirements of each block within a preset time period in the future through a time series prediction model corresponding to a long short-term memory network; and perform detection and feedback on the obtained simulated operation data and the predicted data to adjust the building model so that the data output by the adjusted building model is applied to the modeling of the building central air-conditioning unit.

[0017] At least one of the above technical solutions adopted in the embodiments of the present application can achieve the following beneficial effects: The embodiments of the present application can intuitively represent the structure and layout of the building by constructing an undirected graph of the building architecture, and more accurately determine the connection relationship between the feature points in the building through the graph neural network, providing an accurate basis for subsequent simulation operation and prediction. Through simulation operation, the operating status of the building under different conditions is simulated to provide a reference for actual operation. The long short-term memory network time series prediction model can predict the environmental changes and air-conditioning energy consumption requirements of each block within a preset time period in the future, which is helpful to formulate operation strategies in advance and optimize energy use. Through detection feedback and model adjustment, the performance and accuracy of the building model are continuously optimized to make it closer to the actual operation. Based on the adjusted building model data, the sample data required for accurate building central air-conditioning unit modeling can be obtained, the amount of modeling data can be expanded, and the accuracy of building central air-conditioning unit modeling can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. In the drawings:

[0019] Figure 1 A flow chart of a data construction method for modeling a central air conditioning unit provided in an embodiment of the present application;

[0020] Figure 2 A structural schematic diagram of a data construction device for modeling a central air-conditioning unit provided in an embodiment of the present application.

[0021] Reference numerals:

[0022] 200: data construction device for central air conditioning unit modeling, 201: processor, 202: memory. DETAILED DESCRIPTION

[0023] Embodiments of the present application provide a data construction method, device and medium for modeling a central air conditioning unit.

[0024] In order to enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this application.

[0025] The technical solution proposed in the embodiment of the present invention is described in detail below with reference to the accompanying drawings.

[0026] Figure 1 A flow chart of a data construction method for modeling a central air conditioning unit provided in an embodiment of the present application, such as Figure 1 As shown, the data construction method for central air conditioning unit modeling includes the following steps:

[0027] Step 101: Acquire building structure data, extract feature points from the building structure data, construct an undirected building graph based on the extracted feature points, determine the connection relationship of the feature points corresponding to the undirected building graph through a graph neural network, and construct a building model based on the undirected building graph and the connection relationship of the feature points.

[0028] In one embodiment of the present application, a matching check is performed on multi-source building structure data to filter out abnormal data and obtain reference feature points; the correlation coefficient between the reference feature points is determined by the Spearman rank correlation coefficient, and the correlation coefficient is compared with a preset correlation coefficient threshold to filter out reference feature point pairs that meet the correlation requirements; an undirected building graph is constructed based on the reference feature point pairs. The connection relationship between each reference feature point pair in the building undirected graph is determined by a graph neural network, and the weight between the reference feature point pairs is determined based on the connection relationship; wherein the weight represents the close connection relationship between different nodes. A building model is constructed based on the reference feature points, the connection relationship and the weight.

[0029] Specifically, first, collect building structure data from multiple sources, such as CAD drawings, design documents, building material information, sensor data, etc., and match and verify the data from different sources to ensure the accuracy and consistency of the data. For example, you can check the building location, size, attributes and other information. Identify and filter out abnormal data, which may be caused by sensor failure, data entry errors, etc.

[0030] Furthermore, after the data matching verification, key reference feature points are extracted from the remaining data. These feature points can be important structural elements or functional areas in the building, such as rooms, stairs, elevators, etc., and the Spearman rank correlation coefficient is used to evaluate the correlation between the reference feature points. The calculated correlation coefficient is compared with the preset correlation coefficient threshold, and the embodiment of the present application retains the reference feature point pairs with correlation coefficients higher than the threshold. The selected reference feature points are used as nodes, and an undirected graph of building architecture is constructed according to the correlation between them.

[0031] Furthermore, the connection relationship between each pair of reference feature points in the undirected graph is determined by a graph neural network (GNN), and the complex structural information in the graph is captured by the graph neural network, and the potential connection between the nodes is inferred. Specifically, based on the connection relationship, the weights between the reference feature point pairs are calculated using the graph neural network. Among them, the weight represents the close connection relationship between different nodes. The reference feature points, the connection relationship between the nodes, and the weights are integrated to construct a building model. The building model constructed in the embodiment of the present application can fully reflect the structural characteristics of the building and the interaction between the nodes.

[0032] In one embodiment of the present application, an undirected graph of a building is input into a graph neural network to determine the first relationship score and the connection relationship type between each pair of reference feature points through the graph neural network. Also, the node distances corresponding to each pair of reference feature points are determined, and the number of edges connected to each node is determined. A second relationship score is determined based on the node distance, and a third relationship score is determined based on the number of edges. The first relationship score, the second relationship score, the third relationship score, and the scores corresponding to the connection relationship type are weighted averaged to obtain a comprehensive score. Based on the comprehensive score, the weights between the reference feature point pairs are determined.

[0033] Specifically, an undirected graph of a building structure is taken as input, in which nodes represent reference feature points and edges represent connection relationships between nodes. The input undirected graph is processed by a graph neural network to determine a first relationship score between each pair of reference feature points, wherein the first relationship score is used to represent the correlation or importance between the reference feature point pairs, and the connection relationship type of the reference feature point pairs can also be determined, wherein the connection relationship type can include one of direct connection, indirect connection, strong connection and weak connection.

[0034] Furthermore, by calculating the relative position or proximity of the reference feature point pairs in the graph, the node distance between each reference feature point pair is determined. Also, the number of edges connected to each reference feature point is determined, and the connectivity of the node is determined by the number of connected edges.

[0035] Further, the second relationship score is determined based on the node distance. For example, the closer the node distance is, the higher the second relationship score is, indicating that these reference feature points are closer in space and therefore have a stronger correlation. The third relationship score is determined based on the number of edges. For example, a reference feature point connected to more nodes has a higher importance and is therefore assigned a higher third relationship score.

[0036] Further, the first relationship score, the second relationship score, the third relationship score, and the score corresponding to the connection relationship type are weighted averaged. A comprehensive score is obtained by weighted average processing, and the score comprehensively reflects the various relationship characteristics between each pair of reference feature points. It should be noted that the weight of the weighted average can be set according to actual needs to reflect the importance of different factors in determining the weight, and the embodiment of the present application does not limit this. The higher the weight, the closer the relationship between these reference feature points, and the higher their importance in the building architectural model.

[0037] The embodiment of the present application uses a graph neural network to more comprehensively and accurately determine the weights between reference feature point pairs in an undirected graph of a building, making the subsequently constructed building model more accurate. The embodiment of the present application not only considers the direct connection relationship between nodes, but also considers the spatial proximity and connectivity of nodes, thereby being able to more accurately reflect the structural characteristics of the building and the interaction between nodes.

[0038] Step 102: Acquire initial operating parameters, partition the building model, and match the acquired initial operating parameters to each block.

[0039] In one embodiment of the present application, a subgraph of an undirected graph of a building structure corresponding to each floor in a building construction model is obtained. Each subgraph is input into a preset functional feature extraction model to output a building feature graph of each floor through the preset functional feature extraction model. The building feature graph of each floor is subjected to feature comparison, and based on the comparison result, the building construction model is divided into a plurality of blocks, and based on the building construction structure data corresponding to the subgraph, a block position data set corresponding to each of the plurality of blocks is determined. Based on the block position data set of each floor, the initial operating parameters corresponding to the building construction model are determined; wherein the initial operating parameters include at least one of the initial environmental data, building equipment operation data, building crowd activity data, and weather data corresponding to each block. The initial operating parameters are matched with the block position data set to match the initial operating parameters to each block.

[0040] Specifically, for each floor, a corresponding subgraph is extracted from the undirected graph of the building. These subgraphs contain only nodes and edges within the floor, representing rooms, corridors, etc. and the connection relationship between them. The embodiment of the present application is provided with a convolutional neural network model, which has been trained by a subgraph of the undirected graph of the building corresponding to a large number of annotated floors. The subgraph of the undirected graph of the building corresponding to each floor is input into the convolutional neural network model, and the room type, such as office, conference room, toilet, corridor width, door and window position and other features are identified by the convolutional neural network model, and a building feature map containing these features is generated.

[0041] Furthermore, the building feature graphs of different floors are compared to find out the similarities and differences between them. For example, by comparing the feature graphs of each floor, it is concluded that the first floor is mainly the reception area and public area, while the high floors are mainly offices. Based on the results of feature comparison, the building model is divided into blocks with similar functions and structures. For each block, the building structure data of the block is extracted according to the nodes and edges in the subgraph of the building undirected graph corresponding to the floor. Based on the building structure data, the location data set of each block is determined, wherein the block location data set includes the relative position of the block in the building, the nodes and edges contained, and other information. The initial operating parameters corresponding to the building model are determined. These parameters include environmental data of each block, such as temperature, humidity, operating data of building equipment, such as elevator usage, air conditioning energy consumption, crowd activity data, such as employee commuting time, conference room reservation, and weather data. Different blocks have different functions, so there are certain differences in their initial operating parameters. The initial operating parameters are matched with the block location data set corresponding to each block to ensure that each block is managed and controlled according to the predetermined initial operating parameters.

[0042] In one embodiment of the present application, the block location data set of each floor is combined to generate a floor building structure diagram; the floor building structure diagram is compared with the reference structure diagram in the parameter database to determine a target structure diagram that meets the similarity requirements. Based on the target structure diagram, a required parameter document is determined in the parameter database; wherein the parameter database is provided with multiple reference structure diagrams, and also provided with parameter documents matching the multiple reference structure diagrams.

[0043] For each floor, these block location data sets are combined according to the determined block location data sets to generate a complete floor building structure diagram. Among them, the floor building structure diagram in the embodiment of the present application can be a two-dimensional or three-dimensional plan view or stereogram, which is used to intuitively display the layout and connection relationship of each block in the floor. Suppose a floor has three blocks: Area A (office), Area B (corridor) and Area C (conference room). According to the block location data set, these three blocks are drawn separately on the floor building structure diagram, and the connection relationship between them is indicated (such as Area A is connected to Area C through Area B).

[0044] Furthermore, the embodiment of the present application is pre-set with a parameter database, which stores a large number of floor building structure diagrams and corresponding parameter documents. The generated floor building structure diagram is compared with the reference structure diagram in the parameter database, and based on the comparison result, the target structure diagram that meets the similarity requirements is determined. After the target structure diagram is determined, the parameter document matching the target structure diagram is retrieved from the parameter database. Among them, the parameter document in the embodiment of the present application contains detailed information about the floor building structure, equipment configuration, environmental parameters, crowd activities, etc. For example, assuming that the target structure diagram is an office floor layout, the parameter document matching it contains information such as the office lighting configuration, air conditioning system settings, corridor width and traffic capacity, and conference room audio and video equipment.

[0045] Step 103: Based on the initial operating parameters corresponding to the building model and each block, the building model is simulated and run, and the environmental changes and air-conditioning energy consumption requirements of each block within a preset time period in the future are predicted through the time series prediction model corresponding to the long short-term memory network.

[0046] In one embodiment of the present application, the initial operating parameters are input into the building model, so as to set the initial indoor temperature, humidity, air flow rate and other environmental parameters for each block, and set the initial operating state of the equipment, such as the air conditioning temperature setting value, the lighting switch state, etc., to simulate the operation of the building model. According to the set initial conditions and external conditions, the thermal and humid environment, air flow, energy consumption and other parameters inside the building are calculated. The simulation results are recorded in real time or regularly, including the environmental parameters of each block, the equipment operating state, energy consumption data, etc.

[0047] In one embodiment of the present application, the building structure parameters and equipment placement parameters corresponding to each block are determined, and the physical isolation areas corresponding to each block are determined based on the building structure parameters and the equipment placement parameters. Based on the isolation object information corresponding to the physical isolation area, the first mutual influence parameters corresponding to different physical isolation areas within each block range are determined in the first preset influence parameter database. Based on the distance corresponding to each block and the isolation physical information between the blocks, the second mutual influence parameters corresponding to each block are determined in the second preset influence parameter database. Based on the first mutual influence parameters, a first simulated operation adjustment is performed on the different physical isolation areas within the block range. And, based on the second mutual influence parameters, a second simulated operation adjustment is performed between the blocks after the first adjustment.

[0048] Specifically, the building structure parameters of each block are determined, such as the geometric dimensions of the block, length, width, height, load-bearing structure type, such as frame, floor height, wall material and its thermal performance, etc. And the equipment placement parameters of each block are determined, such as the location, type, power, operating parameters, etc. of various types of equipment placed in the block, such as air-conditioning units, lighting fixtures, distribution boxes, etc. For example, block A is an office area, and its structural parameters include a floor height of 3.6 meters, a lightweight partition wall with a thermal resistance value of R-19, and equipment placement parameters include a central air-conditioning unit installed on the top of block A, each unit has a power of 100kW, and lighting fixtures are evenly distributed on the ceiling, each with a power of 40W.

[0049] Based on the structural parameters and equipment placement parameters of the block, the block is further divided into different physically isolated areas. These areas are separated by walls, floors, ceilings and other isolation objects. For example, block A is divided into four physically isolated areas: the front desk reception area, the open office area, the conference room and the rest area. The front desk reception area and the open office area are separated by glass partitions, and the conference room is separated by solid walls.

[0050] Furthermore, based on the isolation object information corresponding to the physical isolation area, such as material, thickness, thermal resistance, etc., the mutual influence parameters between different physical isolation areas within each block range, such as thermal conductivity, are searched in the first preset influence parameter database. For example, the thermal conductivity coefficient of the glass partition is relatively high, which may cause a greater mutual influence of temperature between the front desk reception area and the open office area. In the first preset influence parameter database, the thermal conductivity influence parameters corresponding to the glass partition can be queried. Based on the distance between each block and the physical information of the isolation between the blocks, such as floor height, wall type, whether adjacent, etc., the mutual influence parameters between the blocks are determined in the second preset influence parameter database.

[0051] Furthermore, based on the first mutual influence parameter, simulated operation adjustment is performed on different physically isolated areas within the block range. For example, the outlet temperature and wind speed of the air conditioning unit are adjusted. Based on the second mutual influence parameter, simulated operation adjustment is performed between the blocks after the first adjustment. Through adjustment, the influence caused by physical isolation within and between blocks is reduced, so that the building model operates in an environment that meets the operation requirements.

[0052] In one embodiment of the present application, simulated operation data corresponding to a building model within a preset time period is obtained. The simulated operation data is input into a time series prediction model corresponding to a long short-term memory network (LSTM) to obtain equipment operation data and environmental change data corresponding to the building model within a future preset time period. An environmental data change line graph is constructed based on the environmental change data, the difference between adjacent data in the environmental data change line graph is determined, and the environmental change section and the environmental stability section are determined based on the difference. The first air-conditioning energy consumption is determined based on the duration of the environmental change section, and the second air-conditioning energy consumption is determined based on the duration of the environmental stability section, so as to obtain an air-conditioning energy consumption demand forecast value based on the first air-conditioning energy consumption and the second air-conditioning energy consumption.

[0053] Specifically, the building model is used to perform simulation operation in a preset time period, such as the past week, month, etc. The simulation operation will generate a large amount of data, including but not limited to the operating status of the equipment, energy consumption data, environmental parameters, etc. The collected simulation operation data is used as input and input into the time series prediction model corresponding to the long short-term memory network. The time series prediction model captures the time dependency in the time series data to obtain the equipment operation data and environmental change data corresponding to the building model in the future preset time period. Among them, the training process of the time series prediction model corresponding to the long short-term memory network is: use historical simulation operation data as training samples to train the model until the model achieves the required prediction performance on the verification set. The trained model is applied to the prediction in the future preset time period to obtain the predicted values ​​of the equipment operation data and environmental change data corresponding to the building.

[0054] Based on the prediction results of the time series prediction model, a line graph of environmental data changes is constructed. By analyzing the line graph of environmental data changes, the difference between adjacent data is determined. Based on these differences, the line graph is divided into environmental change sections, that is, time periods with large changes in parameters such as temperature and humidity, and environmental stability sections, that is, time periods with relatively small changes in parameters.

[0055] Specifically, the air conditioning energy consumption in the environmental change section and the environmental stability section is determined according to the duration of the two sections. In the environmental change section, the air conditioning operation state needs to be constantly adjusted to adapt to environmental changes, so the energy consumption will be relatively high; while in the environmental stability section, the air conditioning can maintain a relatively stable operation state due to small environmental changes, and the energy consumption is relatively low. Assume that through historical data analysis, in the environmental change section, the average air conditioning energy consumption per hour is 10kWh; and in the environmental stability section, the average air conditioning energy consumption per hour is 5kWh. According to the forecast results for the next week, the total duration of the environmental change section is 40 hours, and the total duration of the environmental stability section is 80 hours. Therefore, the energy consumption of the first air conditioner (environmental change section) is 400kWh, and the energy consumption of the second air conditioner (environmental stability section) is 400kWh.

[0056] The first air conditioning energy consumption and the second air conditioning energy consumption are added together to obtain a total predicted value of the air conditioning energy consumption demand in a future preset time period.

[0057] Step 104: Detect and feedback the acquired simulation operation data and prediction data to adjust the building architectural model, so that the data output by the adjusted building architectural model is applied to the modeling of the building central air-conditioning unit.

[0058] In one embodiment of the present application, the duration of the environmental change section is processed by ratio with the duration of the environmental stability section to obtain an environmental change ratio. The first air conditioning energy consumption and the second air conditioning energy consumption are respectively processed by ratio with the predicted value of the air conditioning energy consumption demand to obtain an air conditioning energy consumption ratio. When the environmental change ratio and the air conditioning energy consumption ratio do not meet the preset threshold conditions, a first difference between the environmental change ratio and the preset threshold is determined, and a second difference between the air conditioning energy consumption ratio and the preset threshold is determined. The first difference and the second difference are matched with the model adjustment solution database to obtain a corresponding adjustment solution to adjust the building architectural model.

[0059] Specifically, first calculate the environmental change ratio, that is, the ratio of the duration of the environmental change segment to the duration of the environmental stability segment. This ratio determines the frequency of environmental changes. Assume that the duration of the environmental change segment is 40 hours and the duration of the environmental stability segment is 80 hours. Then the environmental change ratio is 0.5. Next, calculate the air conditioning energy consumption ratio, that is, the ratio of the first air conditioning energy consumption and the second air conditioning energy consumption to the predicted value of the air conditioning energy consumption demand. The distribution of energy consumption in different segments can be determined through this ratio.

[0060] Furthermore, the calculated environment change ratio and air conditioning energy consumption ratio are compared with the preset thresholds to determine whether the current environment change and energy consumption distribution are within a reasonable range. Assuming that the preset environment change ratio threshold is 0.3 to 0.7 and the air conditioning energy consumption ratio threshold is also in a similar range, if the environment change ratio is 0.5 and the air conditioning energy consumption ratio is also 0.5, both meet the preset threshold conditions.

[0061] Furthermore, if the environmental transformation ratio or the air conditioning energy consumption ratio does not meet the preset threshold conditions, the difference between these ratios and the preset thresholds, the first difference and the second difference, are calculated. Then, these differences are matched with the model adjustment solution database to find the most suitable adjustment solution. Assume that the environmental transformation ratio is 0.8 (exceeding the preset upper limit), and the air conditioning energy consumption ratio is still 0.5. The first difference is calculated to be 0.8 - 0.7 = 0.1 (the difference exceeding the upper limit), and this difference is matched with the model adjustment solution database. The database in the embodiment of the present application contains a variety of adjustment solutions, such as adjusting the thermal insulation performance of the building, optimizing the control strategy of the air conditioning system, increasing the use of natural ventilation, etc. According to the matching results, the corresponding adjustment solution is selected and implemented to improve the environmental changes and energy consumption distribution of the building. Finally, according to the matched adjustment solution, the building architectural model is adjusted, and the simulation operation and prediction are re-performed to verify the adjustment effect.

[0062] Based on the adjusted building architectural model, a simulation operation is performed, and the simulation operation data is stored as historical sample data for modeling the building central air-conditioning unit.

[0063] Figure 2 A schematic diagram of a data construction device for modeling a central air conditioning unit provided in an embodiment of the present application. Figure 2As shown, a data construction device 200 for modeling a central air conditioning unit includes: at least one processor 201; and a memory 202 in communication with the at least one processor 201; wherein the memory 202 stores instructions that can be executed by the at least one processor 201, and the instructions are executed by the at least one processor 201, so that the at least one processor 201 can: obtain building structure data, extract feature points from the building structure data, construct a building undirected graph based on the extracted feature points, determine the connection relationship of the feature points corresponding to the building undirected graph through a graph neural network, and The building model is constructed based on the connection relationship of feature points; the initial operating parameters are obtained, the building model is partitioned, and the obtained initial operating parameters are matched to each block; based on the initial operating parameters corresponding to the building model and each block, the building model is simulated and operated, and the environmental changes and air-conditioning energy consumption requirements of each block in the future preset time period are predicted through the time series prediction model corresponding to the long short-term memory network; the obtained simulation operation data and prediction data are tested and fed back to adjust the building model, so that the data output by the adjusted building model can be used for modeling the building central air-conditioning unit.

[0064] A non-volatile computer storage medium provided in an embodiment of the present application stores computer executable instructions, wherein the computer executable instructions are configured to: obtain building structure data, extract feature points from the building structure data, construct an undirected graph of a building based on the extracted feature points, determine the feature point connection relationship corresponding to the undirected graph of the building through a graph neural network, and construct a building model based on the undirected graph of the building and the feature point connection relationship; obtain initial operating parameters, partition the building model, and match the obtained initial operating parameters to each block; simulate the building model based on the initial operating parameters corresponding to the building model and each block, and predict the environmental changes and air-conditioning energy consumption requirements of each block within a preset time period in the future through a time series prediction model corresponding to a long short-term memory network; and perform detection and feedback on the obtained simulated operation data and the predicted data to adjust the building model so that the data output by the adjusted building model is applied to the modeling of the building central air-conditioning unit.

[0065] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0066] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the embodiments of the present application may have various changes and variations. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application.

Claims

1. A data construction method for central air conditioning unit modeling, characterized in that: The method comprises: Acquire building structure data, extract feature points from the building structure data, construct an undirected building graph based on the extracted feature points, determine a connection relationship of the feature points corresponding to the undirected building graph through a graph neural network, and construct a building model based on the undirected building graph and the connection relationship of the feature points; Acquire initial operating parameters, partition the building model, and match the acquired initial operating parameters to each block; Based on the initial operating parameters corresponding to the building model and each block, the building model is simulated and run, and the environmental changes and air conditioning energy consumption requirements of each block in a preset time period in the future are predicted through the time series prediction model corresponding to the long short-term memory network; Performing detection and feedback on the acquired simulation operation data and prediction data to adjust the building architectural model, so that the data output by the adjusted building architectural model is applied to the modeling of the building central air-conditioning unit; The obtaining of initial operating parameters, partitioning the building model, and matching the obtained initial operating parameters to each block specifically includes: Obtaining a subgraph of the building undirected graph corresponding to each floor in the building model; Inputting each of the sub-graphs into a preset functional feature extraction model to output a building feature graph of each floor through the preset functional feature extraction model; Performing feature comparison on the building feature graph of each floor, dividing the building model into a plurality of blocks based on the comparison result, and determining a block position data set corresponding to each of the plurality of blocks based on the building structure data corresponding to the sub-graph; Based on the block location data set of each floor, determining the initial operation parameters corresponding to the building model; wherein the initial operation parameters include at least one of the initial environment data, building equipment operation data, building crowd activity data and weather data corresponding to each block; Matching the initial operating parameters with the block position data set to match the initial operating parameters to each block; The determining of the initial operation parameters corresponding to the building model based on the block position data set of each floor specifically includes: The block location data set of each floor is combined to generate a floor building structure diagram; the floor building structure diagram is compared with a reference structure diagram in a parameter database to determine a target structure diagram that meets the similarity requirements; Based on the target structure diagram, a required parameter document is determined in the parameter database; wherein the parameter database is provided with a plurality of reference structure diagrams, and also provided with parameter documents matching the plurality of reference structure diagrams.

2. A data construction method for central air conditioning unit modeling according to claim 1, characterized in that: The acquiring of building structure data, extracting feature points from the building structure data, constructing an undirected building graph based on the extracted feature points, determining a connection relationship of feature points corresponding to the undirected building graph through a graph neural network, and constructing a building model based on the undirected building graph and the connection relationship of feature points, specifically includes: Performing matching verification on the building structure data from multiple sources to filter out abnormal data and obtain reference feature points; Determine the correlation coefficient between the reference feature points by using the Spearman rank correlation coefficient, and compare the correlation coefficient with a preset correlation coefficient threshold to screen out reference feature point pairs that meet the correlation requirements; Constructing the building undirected graph based on the reference feature point pairs; Determine the connection relationship between each pair of reference feature points in the undirected graph of the building through a graph neural network, and determine the weights between the reference feature point pairs based on the connection relationship; wherein the weights represent the close connection relationship between different nodes; The building model is constructed based on the reference feature points, the connection relationship and the weight.

3. A data construction method for central air conditioning unit modeling according to claim 2, characterized in that: Determining the connection relationship between the reference feature point pairs in the building undirected graph by using a graph neural network, and determining the weights between the reference feature point pairs based on the connection relationship, specifically includes: Inputting the building undirected graph into the graph neural network to determine the first relationship score and the connection relationship type between each pair of reference feature points through the graph neural network; And, determining the node distances corresponding to each pair of reference feature points, and determining the number of edges connected to each node; determining a second relationship score based on the node distance, and determining a third relationship score based on the number of edges; Performing weighted average processing on the first relationship score, the second relationship score, the third relationship score, and the score corresponding to the connection relationship type to obtain a comprehensive score; Based on the comprehensive score, the weights between the reference feature point pairs are determined.

4. A data construction method for central air conditioning unit modeling according to claim 1, characterized in that: After simulating the building model based on the initial operation parameters corresponding to each block, the method further includes: Determine the building structure parameters and equipment placement parameters corresponding to each of the blocks, and determine the physical isolation areas corresponding to each of the blocks based on the building structure parameters and the equipment placement parameters; Based on the isolated object information corresponding to the physically isolated area, determining in the first preset influence parameter database the first mutual influence parameters corresponding to the different physically isolated areas within each of the blocks; Based on the distances respectively corresponding to the blocks and the physical information of the isolation between the blocks, determining the second mutual influence parameters respectively corresponding to the blocks in the second preset influence parameter database; Based on the first mutual influence parameter, performing a first simulation operation adjustment on different physically isolated areas within the block; And, based on the second mutual influence parameter, a second simulation operation adjustment is performed between the blocks after the first adjustment.

5. The data construction method for central air conditioning unit modeling according to claim 1, characterized in that: The time series prediction model corresponding to the long short-term memory network is used to predict the environmental changes and air conditioning energy consumption requirements of each block within a preset time period in the future, specifically including: Acquire simulation operation data corresponding to the building model within a preset time period; Inputting the simulated operation data into the time series prediction model corresponding to the long short-term memory network to obtain the equipment operation data and environmental change data corresponding to the building model in a future preset time period; Building an environmental data change line graph based on the environmental change data, determining the difference between adjacent data in the environmental data change line graph, and determining an environmental change section and an environmental stability section based on the difference; A first air conditioning energy consumption is determined based on the duration of the environmental change section, and a second air conditioning energy consumption is determined based on the duration of the environmental stability section, so as to obtain an air conditioning energy consumption demand forecast value based on the first air conditioning energy consumption and the second air conditioning energy consumption.

6. A data construction method for central air conditioning unit modeling according to claim 5, characterized in that: The detection and feedback of the acquired simulation operation data and prediction data to adjust the building model specifically includes: The duration of the environmental change section is processed by ratio with the duration of the environmental stability section to obtain an environmental change ratio; Ratio processing is performed on the first air conditioning energy consumption and the second air conditioning energy consumption respectively with the air conditioning energy consumption demand forecast value to obtain an air conditioning energy consumption ratio; When the environment change ratio and the air conditioning energy consumption ratio do not meet the preset threshold conditions, determining a first difference between the environment change ratio and the preset threshold, and determining a second difference between the air conditioning energy consumption ratio and the preset threshold; The first difference and the second difference are matched with a model adjustment solution database to obtain a corresponding adjustment solution to adjust the building architectural model.

7. A data construction device for central air conditioning unit modeling, characterized in that: The device comprises a memory for storing computer program instructions and a processor for executing the program instructions, wherein when the computer program instructions are executed by the processor, the device is triggered to execute the method according to any one of claims 1 to 6.

8. A non-volatile computer storage medium storing computer executable instructions, characterized in that: The computer executable instructions can execute the method according to any one of claims 1 to 6.

Images