Method and system for directing undirected graph of building system relationship topology

Abstract

The method and system for directing the topological undirected graph of building system relationships provided by the embodiment of the present invention include: obtaining node information in the topological relationship of the Revit building model; after automatically correcting and filtering the node information, performing data processing on the node information to obtain the adjacency matrix and feature matrix of the node information; using the adjacency matrix and feature matrix as input values, inputting them into a pre-trained feature extraction model for calculation, and obtaining a topological feature vector including node gravity parameters; obtaining the directional relationship between nodes according to the size of the gravity parameter of each node in the topological feature vector, and completing the directing of the undirected graph. The present invention converts undirected graph data lacking information into directed graph relationship connections with information output, and can clearly give the flow direction of water or wind in water pipes or air ducts; it can also check some modeling problems in the Revit model, and can also be extended to other topological data informatization, such as converting a three-dimensional model to a two-dimensional system diagram and its display.

Description

Technical Field

[0001] The present invention relates to the field of building information technology, and in particular to a method and system for directing a building system relationship topology undirected graph. Background Art

[0002] In the building information model, we can extract the connection relationship data between equipment and pipes, but these connection relationships are not well known by people. People's cognition includes the direction of equipment connection (the flow direction of water and wind in the pipe), the connection method of equipment in the system (such as parallel, series, mother connection), and the functional division of equipment (chilled water cycle, cooling water cycle, wind cycle, water cycle). The effective informationization of the system connection of the building can be used to automatically arrange the system diagram from three-dimensional to two-dimensional, and diagnose the correctness of the connection relationship. Summary of the invention

[0003] In order to effectively informatize the building system connection, and then use it to automatically arrange the system diagram from three-dimensional to two-dimensional and diagnose the correctness of the connection relationship, the embodiment of the present invention provides a method and system for directing the topological undirected graph of the building system relationship. The specific technical solution is as follows:

[0004] To achieve the above object, an embodiment of the present invention provides a method for directing an undirected graph of a building system relationship topology, comprising the steps of:

[0005] Get the node information in the topological relationship of the Revit building model;

[0006] After automatically correcting and filtering the node information, data processing is performed on the node information to obtain an adjacency matrix and a characteristic matrix of the node information;

[0007] The adjacency matrix and the feature matrix are used as input values ​​and input into a pre-trained feature extraction model for calculation to obtain a topological feature vector including a gravity parameter for determining the direction of a node; wherein, during the training process of the feature extraction model, the feature matrix of the node information is assigned gravity parameter information;

[0008] The directional relationship between nodes is obtained according to the magnitude of the gravity parameter of each node in the topological feature vector, and the undirected graph is directed.

[0009] Furthermore, the feature extraction model can be trained through the following steps:

[0010] Construct a knowledge base of topological relationships of building mechanical and electrical system equipment;

[0011] Generate node information of the topological relationship;

[0012] Performing data processing on the node information to obtain an adjacency matrix and a feature matrix corresponding to the node information;

[0013] Based on the encoding / decoding framework of the variational autoencoder, the adjacency matrix and the feature matrix are input into the graph convolutional neural network for encoding to generate a latent vector including gravity parameter information, and the latent vector is input into the decoder to obtain the decoded adjacency matrix;

[0014] The decoded adjacency matrix is ​​input into the directed topological loss function for calculation. When the calculation result is less than a preset threshold, the parameters of the graph convolutional neural network are output to obtain a feature extraction model.

[0015] Furthermore, before generating the node information of the topological relationship and after constructing the topological relationship knowledge base of the building mechanical and electrical system equipment, it also includes comparing the Revit model topological data of multiple projects accumulated in the database with the generated topological relationship, and comparing and correcting the generated topological relationship.

[0016] Furthermore, the directed topological loss function includes a topological loss function and a directional loss function. The topological loss function is obtained by calculating the weighted cross entropy by comparing the original adjacency matrix and the decoded adjacency matrix. The directional loss function is:

[0017] Y i，j Yes X i X j Directions

[0018] In the formula, F represents the number of nodes, Z represents the latent vector, X is the feature matrix, the i-th node is connected to the j-th node, and Y ij is 1 if the value is true, otherwise it is 0.

[0019] Furthermore, the node information includes a target node ID, node attributes, other node IDs connected to the target node, and the number of nodes; other node IDs connected to the target node include: other node IDs pointing to the target node and other node IDs pointed to by the target node.

[0020] A second aspect of an embodiment of the present invention further provides a system for converting an undirected graph of a building system relationship topology into a directed system, comprising:

[0021] An acquisition module is used to obtain node information in the topological relationship of the Revit building model;

[0022] A processing module, used for automatically correcting and filtering the node information, and then performing data processing on the node information to obtain an adjacency matrix and a feature matrix of the node information;

[0023] A calculation module, used for inputting the adjacency matrix and the feature matrix as input values ​​into a pre-trained feature extraction model for calculation to obtain a topological feature vector including a node gravity parameter;

[0024] The undirected graph directed module is used to obtain the directional relationship between nodes according to the magnitude of the gravity parameter of each node in the topological feature vector, and complete the directed transformation of the undirected graph.

[0025] The third aspect of an embodiment of the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the processor processes the steps of the method for directing an undirected graph of a building system relationship topology as described above.

[0026] A fourth aspect of the present invention provides an electronic device, the electronic device comprising:

[0027] processor; and,

[0028] A memory is arranged to store computer executable instructions, which, when executed, cause the processor to perform the method for directing the undirected graph of building system relationship topology.

[0029] The method for directing a topological undirected graph of a building system relationship provided in an embodiment of the present invention includes: obtaining node information in the topological relationship of a Revit building model; after automatically correcting and filtering the node information, performing data processing on the node information to obtain an adjacency matrix and a feature matrix of the node information; using the adjacency matrix and the feature matrix as input values, inputting them into a pre-trained feature extraction model for calculation to obtain a topological feature vector including node gravity parameters; obtaining the directional relationship between nodes according to the gravity parameter of each node in the topological feature vector, and completing the directing of the undirected graph. The present invention converts data lacking information into a directed relationship connection with information output, and can clearly give the flow direction of water or wind in a water pipe or air duct; it can also check some modeling problems in the Revit model, and can also be extended to other topological data informatization, such as converting a three-dimensional model to a two-dimensional system diagram and its display. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 A flowchart of a method for directing an undirected graph of building system relationship topology provided in Example 1 of the present invention;

[0031] Figure 2 A structural block diagram of a system for converting an undirected graph of a building system relationship topology into a directed one provided by Embodiment 1 of the present invention;

[0032] Figure 3An example of a basic connection graph of a method for directing an undirected graph of a building system relationship topology provided in Embodiment 1 of the present invention;

[0033] Figure 4 for Figure 3 Schematic diagram of the connection between the primary return air valve and the secondary return air valve. DETAILED DESCRIPTION

[0034] In order to clearly and comprehensively present the technical solution of the present invention, the present invention is described below in conjunction with the accompanying drawings, but the scope of the present invention is not limited thereto.

[0035] See also Figure 1 A flowchart of a method for directing an undirected graph of a building system relationship topology provided in Example 1 of the present invention includes the following steps:

[0036] S1 obtains node information in the topological relationship of the Revit building model base diagram.

[0037] The above-mentioned nodes include target nodes and other nodes; the above-mentioned node information includes target node ID, node attributes, other node IDs connected to the target node and the number of nodes; other node IDs connected to the target node include: other node IDs pointing to the target node and other node IDs pointed to by the target node; in an optional embodiment of the present invention, the above-mentioned node information is represented in the form of a multidimensional array, as shown below: [

[0039] [0,10,1,[7],[1,3,]],

[0040] [1,5,2,[0,2],[0,2]],

[0041] [2,10,4,[1,4],[5,7]],

[0042] [3,2,2,[0],[4]],

[0043] [4,3,2,[3],[2]],

[0044] [5,3,2,[2],[6]],

[0045] [6,18,1,[5],[]],

[0046] [7,9,2,[2],[0]], ]

[0048] The number of rows in the above multidimensional array indicates how many nodes there are in the topological relationship. The meanings of the elements in each row are as follows:

[0049] The 0th bit represents the ID bit; the 1st bit represents the device type; the 2nd bit represents the number of connection points; the 3rd bit represents the node ID pointed to; and the 4th bit represents the node ID pointing to this node.

[0050] The above node information is stored in the corresponding files of the computer, including connection relationship files, equipment files, waterway files, pipeline files, and connection object files. Among them, the connection relationship includes the ID of each point, the ID of the point connected to the point, the ID of the object to which the point belongs, the type of the object (equipment, waterway, pipeline, object), etc. According to the ID of the object, the attribute information of the object can be found in the corresponding type file, including: position, outline, model, type code, etc.

[0051] S2 performs data processing on the node information after automatically correcting and filtering the node information to obtain an adjacency matrix and a feature matrix of the node information.

[0052] Since in the present invention, the undirected graph is changed into a directed graph, most of it only involves the connection relationship between devices, and there is no need to reflect the pipeline. In the Revit building model, it is generally modeled according to different floors. Therefore, it is necessary to simplify the topology diagram, only retain the equipment and valves, directly connect all the equipment or valves connected by water pipes, air ducts, and connectors, and delete the water pipes, air ducts, and connectors in between.

[0053] Delete individual equipment or valves without topological connections in the Revit building model.

[0054] An optional implementation of the embodiment of the present invention also includes checking whether the cross sections of the pipe ports are parallel and whether the port distance is within 10 mm, and modifying the topological data to connect the pipes.

[0055] The adjacency matrix is ​​a matrix that represents the neighbor relationship between nodes, and the feature matrix is ​​composed of device attributes, where the device attributes include: device type, device location, device profile, and type code, etc. The adjacency matrix is ​​obtained by data processing of node information.

[0056] S3 uses the adjacency matrix and the feature matrix as input values ​​and inputs them into a pre-trained feature extraction model for calculation to obtain a topological feature vector including node gravity parameters.

[0057] The above feature extraction model can be trained through the following steps:

[0058] S31 constructs a knowledge base of topological relationships of building electromechanical system equipment, where each topological relationship can be represented as a basic connection diagram and an extensible rule;

[0059] A large amount of topological relationship data is generated according to the extensible rules; based on the extensible rules, an extensible object is listed, which can be a node or a group of nodes. The extensible node list is expressed as follows: [[7], [5, 6]], where node 7 can be expanded and added to the graph individually, and nodes 5 and 6 can be expanded and added to the graph as a group. Other constraints, such as a certain type of node being smaller than another type of node, are screened in the topology generation program corresponding to different types of topological relationships.

[0060] The above basic connection diagram and scalable rules are as follows:

[0061] See also Figure 3 , if it is a topological relationship diagram of the air conditioning system (air conditioning unit) on the user side, in the figure, the nodes of its basic connection diagram include fresh air inlet, air conditioning unit, supply air outlet, return air outlet, and exhaust outlet; among them, the return air outlet node and the fresh air outlet node point to the air conditioning unit node respectively, and the air conditioning unit node points to the supply air outlet node and the exhaust air outlet node respectively.

[0062] In order to expand the topological relationship, an expansion rule can be set to expand the expert library. The rule is as follows: A) (1) (2) can be connected in parallel, (1) quantity is less than (2) quantity; B) (3) the return air outlet has or does not have a primary or secondary return air valve. This rule can be found in Figure 4 ; C) with or without the combination of (1) and (3); D) (4) the combination relationship cannot be connected in parallel, etc.

[0063] S32 generates node information of the topological relationship.

[0064] A multidimensional array of a basic graph structure is generated according to the basic connection graph, and the multidimensional array is used to express node information. The node information includes the target node id, node attributes, other node ids connected to the target node, and the number of nodes; other node ids connected to the target node include: other node ids pointing to the target node and other node ids pointed to by the target node; in an optional embodiment of the present invention, the above node information is represented in the form of a multidimensional array. The array is the same as the multidimensional array in the model learning process.

[0065] S33 performs data processing on the node information to obtain an adjacency matrix and a feature matrix corresponding to the node information; the adjacency matrix and the feature matrix are the same as those in the model learning process.

[0066] S34 inputs the adjacency matrix and the feature matrix into a graph convolutional neural network model trained and optimized under the encoding / decoding framework of a variational autoencoder for encoding, generates a latent vector including gravity parameter information, and then inputs the obtained latent vector into a decoder for decoding to obtain a decoded adjacency matrix.

[0067] The above-mentioned feature vector is a vector including one-dimensional gravity parameter information. The encoder in the encoding / decoding framework of the above-mentioned variational autoencoder adopts a graph convolutional neural network as an encoder for encoding. The calculation process of the above-mentioned encoder and decoder belongs to the existing algorithm in this field, and its specific calculation process is not repeated here.

[0068] S35 inputs the decoded adjacency matrix into the directed topological loss function for calculation. When the calculation result of the topological feature vector is less than a preset threshold, the parameters of the graph convolutional neural network are output to obtain a feature extraction model.

[0069] The above-mentioned topological feature vector is a multidimensional feature vector including the gravity parameter feature parameters of the nodes. The orientation between the nodes can be determined by comparing the magnitudes of the gravity parameters in the topological feature vector; the above-mentioned directed topological loss function includes a direction loss function and a topological loss function.

[0070] In an optional implementation of an embodiment of the present invention, before generating the node information of the topological relationship and after constructing the topological relationship knowledge base of the building mechanical and electrical system equipment, it also includes comparing the Revit model topological data of multiple projects accumulated in the database with the generated topological relationship, and comparing and correcting the generated topological relationship.

[0071] Preferably, in the embodiment of the present invention, corresponding corrections are mainly made mainly for some frequently missing situations in the actual topology.

[0072] The method for directing a topological undirected graph of a building system relationship provided in an embodiment of the present invention includes: obtaining node information in the topological relationship of a Revit building model; after automatically correcting and filtering the node information, performing data processing on the node information to obtain an adjacency matrix and a feature matrix of the node information; using the adjacency matrix and the feature matrix as input values, inputting them into a pre-trained feature extraction model for calculation to obtain a topological feature vector including node gravity parameters; obtaining the directional relationship between nodes according to the gravity parameter of each node in the topological feature vector, and completing the directing of the undirected graph. The present invention converts data lacking information into a directed relationship connection with information output, and can clearly give the flow direction of water or wind in a water pipe or air duct; it can also check some modeling problems in the Revit model, and can also be extended to other topological data informatization, such as converting a three-dimensional model to a two-dimensional system diagram and its display.

[0073] A second aspect of the embodiments of the present invention further provides a system for converting an undirected graph of building system relationship topology into a directed system. Figure 2 A structural block diagram of a system for converting an undirected graph of a building system relationship topology into a directed system according to Embodiment 1 of the present invention includes:

[0074] An acquisition module is used to obtain node information in the topological relationship of the Revit building model;

[0075] A processing module, used for automatically correcting and filtering the node information, and then performing data processing on the node information to obtain an adjacency matrix and a feature matrix of the node information;

[0076] A calculation module, used for inputting the adjacency matrix and the feature matrix as input values ​​into a pre-trained feature extraction model for calculation to obtain a topological feature vector including a node gravity parameter;

[0077] The undirected graph is directed, and is used to obtain the directional relationship between nodes according to the magnitude of the gravity parameter of each node in the topological feature vector, so as to complete the directed transformation of the undirected graph.

[0078] The third aspect of an embodiment of the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the processor processes the steps of the method for directing an undirected graph of a building system relationship topology as described above.

[0079] A fourth aspect of the present invention provides an electronic device, the electronic device comprising:

[0080] processor; and,

[0081] A memory is arranged to store computer executable instructions, which, when executed, cause the processor to perform the method for directing the undirected graph of building system relationship topology.

[0082] Although the present invention has been described in detail above by general description and specific embodiments, it is obvious to those skilled in the art that some modifications or improvements can be made to the present invention. Therefore, these modifications or improvements made without departing from the spirit of the present invention all belong to the scope of protection claimed by the present invention.

Claims

1. A method for directing an undirected graph of building system relationship topology, characterized in that: Includes steps: Get the node information in the topological relationship of the Revit building model; After automatically correcting and filtering the node information, data processing is performed on the node information to obtain an adjacency matrix and a characteristic matrix of the node information; The adjacency matrix and the feature matrix are used as input values ​​and input into a pre-trained feature extraction model for calculation to obtain a topological feature vector including a node gravity parameter; wherein, during the training process of the feature extraction model, the feature matrix of the node information is assigned gravity parameter information; the orientation between nodes is determined by comparing the size of the gravity parameter in the topological feature vector; According to the magnitude of the gravity parameter of each node in the topological feature vector, the directional relationship between the nodes is obtained to complete the directed transformation of the undirected graph; The feature extraction model can be trained by the following steps: Construct a knowledge base of topological relationships of building mechanical and electrical system equipment; Generate node information of the topological relationship; Performing data processing on the node information to obtain an adjacency matrix and a feature matrix corresponding to the node information; Based on the encoding / decoding framework of the variational autoencoder, the adjacency matrix and the feature matrix are input into a graph convolutional neural network for encoding to generate a latent vector including gravity parameter information, and then the latent vector is input into a decoder of the encoding / decoding framework of the variational autoencoder to obtain a decoded adjacency matrix; Inputting the decoded adjacency matrix into a directed topological loss function for calculation, and when the calculation result is less than a preset threshold, outputting the parameters of the graph convolutional neural network to obtain a feature extraction model; The directed topological loss function includes a topological loss function and a directional loss function. The topological loss function is obtained by calculating the weighted cross entropy by comparing the original adjacency matrix and the decoded adjacency matrix. The directional loss function is: Y i,j Yes X i X j Directions In the formula, F represents the number of nodes, Z represents the latent vector, X is the feature matrix, the i-th node is connected to the j-th node, and Y i,j is 1 if the value is true, otherwise it is 0.

2. The method for converting an undirected graph of building system relationship topology into a directed graph according to claim 1, characterized in that: Before generating the node information of the topological relationship and after constructing the topological relationship knowledge base of the building mechanical and electrical system equipment, it also includes comparing the Revit model topological data of multiple projects accumulated in the database with the generated topological relationship, and comparing and correcting the generated topological relationship.

3. The method for converting an undirected graph of building system relationship topology into a directed graph according to claim 1, characterized in that: The node information includes the target node ID, node attributes, other node IDs connected to the target node, and the number of nodes; The IDs of other nodes connected to the target node include: the IDs of other nodes pointing to the target node and the IDs of other nodes pointed to by the target node.

4. A directed system for topological undirected graph of building system relations, characterized in that: include: An acquisition module is used to obtain node information in the topological relationship of the Revit building model; A processing module, used for automatically correcting and filtering the node information, and then performing data processing on the node information to obtain an adjacency matrix and a feature matrix of the node information; A calculation module, used to input the adjacency matrix and the feature matrix as input values ​​into a pre-trained feature extraction model for calculation, and obtain a topological feature vector including node gravity parameter features; wherein, during the training process of the feature extraction model, the feature matrix of the node information is assigned gravity parameter information; the orientation between nodes is determined by comparing the size of the gravity parameter in the topological feature vector; An undirected graph directed module, used to obtain the directional relationship between nodes according to the magnitude of the gravity parameter of each node in the topological feature vector, and complete the directed transformation of the undirected graph; The feature extraction model can be trained by the following steps: Construct a knowledge base of topological relationships of building mechanical and electrical system equipment; Generate node information of the topological relationship; Performing data processing on the node information to obtain an adjacency matrix and a feature matrix corresponding to the node information; Based on the encoding / decoding framework of the variational autoencoder, the adjacency matrix and the feature matrix are input into a graph convolutional neural network for encoding to generate a latent vector including gravity parameter information, and then the latent vector is input into a decoder of the encoding / decoding framework of the variational autoencoder to obtain a decoded adjacency matrix; Inputting the decoded adjacency matrix into a directed topological loss function for calculation, and when the calculation result is less than a preset threshold, outputting the parameters of the graph convolutional neural network to obtain a feature extraction model; The directed topological loss function includes a topological loss function and a directional loss function. The topological loss function is obtained by calculating the weighted cross entropy by comparing the original adjacency matrix and the decoded adjacency matrix. The directional loss function is: Y i,j Yes X i X j Directions In the formula, F represents the number of nodes, Z represents the latent vector, X is the feature matrix, the i-th node is connected to the j-th node, and Y i,j is 1 if the value is true, otherwise it is 0.

5. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor processes the steps of the method for directing an undirected graph of a building system relationship topology.

6. An electronic device, characterized in that: The electronic device includes: processor; and, A memory is arranged to store computer executable instructions, which, when executed, cause the processor to perform the method for directing the undirected graph of building system relationship topology.

Images