JP7620287B2 - Method and program for detecting piping design defects, method and program for constructing a database of past piping design defects, and system for detecting piping design defects - Google Patents

Description

The present invention relates to a method and program for detecting piping design defects by referring to past defects, a method and program for constructing a database of past piping design defects, and a system for detecting piping design defects.

In recent years, design lead times have tended to shorten, and automation in design work has become increasingly important. Especially in shipbuilding, where a wide variety of products are produced in small quantities, there are many new designs. Furthermore, piping design, which is an outfitting item, is often put off compared to hull design, making this an urgent issue. On the other hand, some automated functions, such as automatic interference checking, are becoming more and more common as CAD functions.

Here, Patent Document 1 discloses a plant equipment layout design system that performs equipment layout design for a multi-story plant in three dimensions using 3D CAD, which includes a standard layout rule database that predetermines the positions in three-dimensional space of the equipment that makes up the plant, inputs basic plant plan information such as plant output, height of each floor, equipment configuration, and building dimensions into a calculation processing device, sets a reference coordinate system in three-dimensional space based on information on standard structural materials included in the basic plant plan information, and sets the virtual space of equipment, piping, passageways, and each floor on the coordinate system as equipment layout positions in accordance with the layout rules to perform the equipment layout design.
Patent Document 2 also discloses a system for designing a building equipment system, the system having at least one processor operative to execute at least one computer program, and at least one user interface connected to the at least one processor and operative to receive a plurality of data sets, the plurality of data sets including a first data set having locations of a plurality of component devices of the at least one building equipment system and a second data set having at least one operational parameter associated with at least one of the plurality of component devices, the at least one computer program uses the plurality of data sets to generate a plurality of design solutions, each of the plurality of design solutions having an arrangement of interconnections between the plurality of component devices, and the at least one computer program operative to determine a construction cost associated with each of the plurality of design solutions.
Patent Document 3 discloses a simulation device that creates a piping model, which is a model of a mesh piping circuit into which a fluid flows, based on an input mesh coarseness, the piping model having a plurality of valves including a plurality of solenoid valves that can be opened and closed by control, and a plurality of pipes, the plurality of pipes being arranged in a mesh shape by connecting the valves of each of the plurality of pipes, and creates a factory model, which is a model of a factory equipped with a mesh piping circuit and is a model to be simulated, by combining the created piping model with a factory layout model indicating a model of a factory using the mesh piping circuit; a construction cost calculation unit that calculates the construction cost of the factory indicated by the factory model; a production cost calculation unit that calculates the operating cost of the factory using the mesh piping circuit by simulating the factory model; and a mesh effect evaluation unit that evaluates the effect of the mesh piping circuit based on the construction cost and operating cost.
Furthermore, Patent Document 4 discloses a design support device that performs a design rule check on design data of a board on which circuit elements are arranged, the device including a first DRC unit that performs a design rule check on the design data by geometric calculation, a generation unit that generates image data from the design data in which a portion determined to be an error by the first DRC unit is centered, a second DRC unit that performs a design rule check on the image data by artificial intelligence that has undergone deep learning, and a display control unit that causes a display unit to display first information indicating the result of the design rule check by the second DRC unit.
Furthermore, Patent Document 5 discloses a piping transport route search method in which, when transporting a fluid from its source to its destination, a computer searches for an optimal transport route from multiple piping systems between the source and destination. The method regards all system equipment that constitutes the piping system, except for valves and piping, as nodes in graph theory, and regards the valves and piping as lines connecting the nodes, thereby disclosing a network representation method for a piping system in which an actual piping system is represented as a network for transport route search.

JP 2006-330887 A Special Publication No. 2013-516007 International Publication No. 2020/183647 WO 2020/79810 Japanese Patent Application Publication No. 7-225788

Patent Document 1 describes a system in which rules for the placement of equipment on each floor are established in advance based on past cases and the knowledge of experienced engineers, and equipment and the like are automatically placed according to those placement rules. However, it does not detect design defects in piping.
Although US Pat. No. 5,999,633 determines a minimum cost solution that satisfies all necessary design parameters input into the system, it does not detect piping design flaws.
Patent Document 3 evaluates the effectiveness of a mesh piping circuit based on construction costs and operating costs, but does not detect design defects in the piping.
Patent Document 4 aims to increase the efficiency of the task of checking whether an error determined by a design rule check is a true error, but does not detect design defects in piping.
Although Patent Document 5 discloses a method for searching for an optimal transfer route by a computer, it does not detect design defects in piping.
Therefore, an object of the present invention is to provide a piping design defect detection method, a detection program, a method for building a past piping design defect database, a construction program, and a piping design defect detection system that quickly detect defects in pipe connections or flow direction at pipe junctions.

A piping design defect detection method corresponding to the description of claim 1 is a method for detecting piping design defects by referring to past defects, and is characterized by comprising a CAD data acquisition step for acquiring design CAD data related to the piping design, a system diagram data acquisition step for acquiring system diagram data of the piping, a directed graph conversion step for converting into a directed graph based on the design CAD data and the system diagram data, a defect information acquisition step for acquiring a past defect graph from a past piping design defect database, a candidate group generation step for generating a candidate group by subgraph isomorphism of the directed graph for the acquired past defect graph, a similarity calculation step for calculating the similarity between the candidate group and the past defect graph, and a ranking presentation step for arranging the candidate group and the past defect graph in order of similarity and presenting a ranking.
According to the present invention described in claim 1, by obtaining past similar defect information ranked based on similarity, it is possible to quickly detect piping design defects, such as pipes that are not connected or flow in the opposite direction at a pipe junction.
In addition, when the design CAD data and the piping system diagram data are managed in an integrated manner in association with each other, it is also possible to integrate the CAD data acquisition step and the system diagram data acquisition step.

The present invention as described in claim 2 is characterized in that the design CAD data includes piping data expressed in line form and the coordinate values of the smallest rectangular parallelepiped of the shape parts, and the system diagram data includes the start and finish conditions of the piping flow.
According to the present invention as set forth in claim 2, by using design CAD data and system diagram data that satisfy the necessary conditions for conversion into a directed graph, the design CAD data and system diagram data can be converted into a directed graph with high accuracy.

The present invention as described in claim 3 is characterized in that the directed graph conversion step includes a definition step for defining graph nodes based on line-format piping data and rectangular parallelepiped coordinate value data, and defining edges connecting the nodes from the piping data expressed in line format, a contraction step for contracting the nodes, and a directed graph construction step for searching for nodes based on start and goal conditions, and constructing a directed graph by assigning the shortest path and flow direction.
According to the third aspect of the present invention, it is possible to convert design CAD data and system diagram data into a directed graph efficiently and with high accuracy.

The present invention as described in claim 4 is characterized in that the contraction step includes a first contraction step for contracting nodes for a point group represented by all nodes, taking into account a specified radius range, and a second contraction step for contracting the nodes of the shape parts and the other internal point nodes into one, taking into account other points inside the rectangular parallelepiped.
According to the fourth aspect of the present invention, the conversion to a directed graph can be more appropriately performed by contracting.

The present invention as set forth in claim 5 is characterized in that a design defect of a piping is corrected based on the past defect graph presented as a ranking in the ranking presentation step.
According to the present invention as defined in claim 5, design defects in piping can be appropriately corrected.
In addition, piping design defects can be corrected either manually or automatically using a computer, CAD system, etc.

A piping design defect detection program corresponding to the description of claim 6 is a program for detecting piping design defects by referring to past defects, and is characterized in that it causes a computer to execute a CAD data acquisition step, a system diagram data acquisition step, a directed graph conversion step, a defect information acquisition step, a candidate group generation step, a similarity calculation step, and a ranking presentation step in a piping design defect detection method.
According to the present invention as set forth in claim 6, a program causes a computer to execute the detection of piping design defects by referring to past defects, thereby making it possible to quickly detect piping design defects and reduce labor.

A method for constructing a past piping design defect database corresponding to the method described in claim 7 is a method for constructing past piping design defects as a database, and is characterized in that it includes a design defect information acquisition step for acquiring information on past piping design defects, a past CAD data acquisition step for acquiring past design CAD data related to the past piping design defects, a past system diagram data acquisition step for acquiring past system diagram data related to the past piping design defects, a past design directed graph conversion step for converting the past piping design defects into directed graphs based on the past design CAD data and the past system diagram data, and a storage step for storing the directed graph-converted design defects in the past piping design defect database as past defect graphs.
According to the present invention as set forth in claim 7, by constructing a database in which past design CAD data and past system diagram data in which there have been past piping design defects are appropriately converted into directed graphs and stored, defects in the current piping design can be quickly identified based on the similarity of the directed graphs.

The present invention as described in claim 8 is characterized in that the past design CAD data includes piping data expressed in line form and coordinate values of the smallest rectangular parallelepiped of the shape parts, and the past system diagram data includes start and finish conditions of the past piping flow.
According to the present invention as set forth in claim 8, by using past design CAD data and past system diagram data that satisfy the necessary conditions for conversion into a directed graph, the past design CAD data and the past system diagram data can be converted into a directed graph with high accuracy.

The present invention as described in claim 9 is characterized in that the past design directed graph conversion step includes a definition step of defining graph nodes based on line-format piping data and rectangular parallelepiped coordinate value data, and defining edges connecting the nodes from the piping data expressed in line format, a contraction step of contracting the nodes, and a directed graph construction step of searching for nodes based on start and goal conditions, and constructing a directed graph by assigning the shortest path and flow direction.
According to the ninth aspect of the present invention, previous design CAD data and system diagram data can be converted into a directed graph efficiently and with high accuracy.

The present invention as described in claim 10 is characterized in that the contraction step includes a first contraction step of contracting nodes for a point group represented by all nodes, taking into account a specified radius range, and a second contraction step of contracting the nodes of the shape parts and other internal point nodes into one, taking into account other points inside the rectangular prism.
According to the tenth aspect of the present invention, the conversion to a directed graph can be more appropriately performed by contracting.

A past piping design defect database construction program corresponding to the claim 11 is a program for constructing past piping design defects as a database, and is characterized in that it causes a computer to execute a design defect information acquisition step, a past CAD data acquisition step, a past system diagram data acquisition step, a past design directed graph conversion step, and a storage step in a past piping design defect database construction method.
According to the present invention as set forth in claim 11, a program causes a computer to execute the construction of a database relating to past piping design defects, thereby making it possible to reduce the labor required for constructing a database that accumulates past piping design defects.

A piping design defect detection system corresponding to the claim 12 is a detection system that detects piping design defects by referring to past defects, and is characterized in comprising defect detection means having a design means for designing piping using a CAD system, a past piping design defect database in which defect information of past piping designs is stored as a past defect graph, a design information acquisition unit for acquiring design CAD data and system diagram data of piping designed by the design means, a past defect information acquisition unit for acquiring defect information of past piping designs in the past piping design defect database, a similarity search unit for converting the design CAD data and the system diagram data into a directed graph, calculating a similarity with the defect information of the past piping designs, and calculating a ranking, and a result presentation unit for presenting the ranking.
According to the present invention described in claim 12, by obtaining past similar defect information ranked based on similarity, it is possible to quickly detect piping design defects, such as pipes that are not connected or flow in the opposite direction at a pipe junction.

In a thirteenth aspect of the present invention, the similarity search section includes a directed graph conversion section, a candidate group generation section, a similarity calculation section, and a ranking calculation section.
According to the present invention as set forth in claim 13, it is possible to accurately calculate the similarity between the directed graph of the current piping design and the directed graph of the past piping design, and perform a ranking calculation.

The present invention as set forth in claim 14 is characterized in that the past piping design defect database is constructed based on a method for constructing a past piping design defect database.
According to the present invention as set forth in claim 14, by constructing a database in which past design CAD data and past system diagram data in which there have been past piping design defects are appropriately converted into directed graphs and stored, defects in the current piping design can be quickly identified based on the similarity of the directed graphs.

The present invention as set forth in claim 15 is characterized in that the defect detection means is utilized in a method for detecting defects in piping design.
According to the fifteenth aspect of the present invention, design defects in piping can be detected quickly and accurately.

The present invention as set forth in claim 16 is characterized in that the rankings of the result presentation section are displayed on the design means together with related information.
According to the present invention as set forth in claim 16, designers and the like can appropriately correct piping design defects by referring to highly similar defects and related information from the similarity results presented in a ranking format.

The present invention as set forth in claim 17 is characterized in that a design means, a past piping design defect database, and a defect detection means are arbitrarily combined and information is linked by utilizing an information communication network.
According to the present invention as described in claim 17, it is possible for one company to distribute the various means and databases to multiple locations and use them, or for a design company to install design means and defect detection means and a database provider to install only a database of past piping design defects, so that each company can use the system according to their respective business operations.

The method for detecting piping design defects of the present invention provides past similar defect information ranked based on similarity, making it possible to quickly detect piping design defects such as unconnected pipes or reversed flow at a pipe junction.

In addition, if the design CAD data contains piping data expressed in line format and the coordinate values of the smallest rectangular parallelepiped of the shape parts, and the system diagram data contains the start and finish conditions of the piping flow, the design CAD data and system diagram data that meet the necessary conditions for converting to a directed graph can be used to accurately convert the design CAD data and system diagram data into a directed graph.

In addition, if the directed graph conversion step includes a definition step that defines graph nodes from line-format piping data and rectangular parallelepiped coordinate data, and defines edges connecting the nodes from the line-format piping data, a contraction step that contracts the nodes, and a directed graph construction step that searches for nodes based on start and goal conditions, and assigns the shortest path and flow direction to construct a directed graph, the design CAD data and system diagram data can be converted into a directed graph efficiently and with even greater accuracy.

In addition, if the contraction step includes a first contraction step that contracts the nodes for the point group represented by all the nodes, taking into account the range of the specified radius, and a second contraction step that contracts the nodes of the shape parts and the other internal point nodes into one, taking into account other points inside the rectangular parallelepiped, the conversion to a directed graph can be performed more appropriately by contracting.

In addition, when correcting a piping design defect based on the past defect graphs presented as rankings in the ranking presentation step, the piping design defect can be appropriately corrected.

In addition, according to the piping design defect detection program of the present invention, the program causes a computer to detect piping design defects by referencing past defects, thereby quickly detecting piping design defects and reducing labor.

In addition, according to the method of constructing a database of past piping design defects of the present invention, by constructing a database in which past design CAD data and past system diagram data in which there have been past piping design defects are appropriately converted into directed graphs and stored, defects in the current piping design can be quickly identified based on the similarity of the directed graphs.

In addition, if the past design CAD data contains piping data expressed in line format and the coordinate values of the smallest rectangular parallelepiped of the shape parts, and the past system diagram data contains the start and finish conditions of the past piping flow, then by using the past design CAD data and past system diagram data that meet the necessary conditions for converting to a directed graph, the past design CAD data and past system diagram data can be converted to a directed graph with high accuracy.

In addition, if the past design directed graph conversion step includes a definition step that defines graph nodes from line-format piping data and rectangular parallelepiped coordinate value data, and defines edges connecting the nodes from the line-format piping data, a contraction step that contracts the nodes, and a directed graph construction step that searches for nodes based on start and goal conditions, and assigns the shortest path and flow direction to construct a directed graph, the past design CAD data and system diagram data can be converted into a directed graph efficiently and with even greater accuracy.

In addition, if the contraction step includes a first contraction step that contracts the nodes for the point group represented by all the nodes, taking into account the range of the specified radius, and a second contraction step that contracts the nodes of the shape parts and the other internal point nodes into one, taking into account other points inside the rectangular parallelepiped, the conversion to a directed graph can be performed more appropriately by contracting.

In addition, according to the past piping design defect database construction program of the present invention, the program causes a computer to construct a database related to past piping design defects, thereby reducing the labor required to construct a database that accumulates past piping design defects.

In addition, the piping design defect detection system of the present invention can obtain past similar defect information ranked based on similarity, making it possible to quickly detect piping design defects, such as pipes that are not connected or flow in the opposite direction at a pipe junction.

In addition, if the similarity search unit has a directed graph conversion unit, a candidate group generation unit, a similarity calculation unit, and a ranking calculation unit, it can accurately calculate the similarity between the directed graph of the current piping design and the directed graph of the past piping design, and perform ranking calculations.

In addition, if the past piping design defect database is constructed based on the past piping design defect database construction method, by constructing a database in which past design CAD data and past system diagram data in which there have been past piping design defects are appropriately converted into directed graphs and stored, current piping design defects can be quickly identified based on the similarity of the directed graphs.

In addition, when the defect detection means is used in a method for detecting piping design defects, it can quickly and accurately detect piping design defects.

In addition, if the rankings from the result presentation section are displayed on the design means together with related information, designers etc. can refer to the similarity results presented in ranking format and the highly similar defects and related information to appropriately correct piping design defects.

In addition, when design means, past piping design defect databases, and defect detection means are arbitrarily combined and information is linked using an information and communication network, it becomes possible for one company to distribute the means and databases to multiple locations and use them, or for a design company to install design means and defect detection means and a database provider to install only the past piping design defect database, and for each company to use the system according to their respective business operations.

1 is a schematic diagram of a method for detecting piping design flaws according to an embodiment of the present invention; Flow chart of the method for detecting piping design defects A schematic diagram of how the non-graph structure data of the same design CAD data is converted into directed graph structure data Flow diagram of the conversion method to the directed graph FIG. 1 shows an example of data stored in the past piping design defect database. Diagram of the system configuration for detecting piping design defects FIG. 1 shows results obtained with an embodiment of the present invention.

This article describes a method and program for detecting piping design defects, a method and program for building a database of past piping design defects, and a system for detecting piping design defects according to an embodiment of the present invention.

Fig. 1 is a schematic diagram of a method for detecting piping design defects according to the present embodiment. Fig. 2 is a flow diagram of the method for detecting piping design defects, in which Fig. 2(a) shows a design correction flow by subgraph isomorphic similarity search, and Fig. 2(b) shows a flow for building a past piping design defect database. Fig. 3 is a schematic diagram of a process in which non-graph structure data of design CAD data is converted into directed graph structure data.
The piping design defect detection method is a method for detecting piping design defects by referring to past defects, and includes a CAD data acquisition step S1 for acquiring design CAD data related to the piping design, a system diagram data acquisition step S2 for acquiring system diagram data of the piping, a directed graph conversion step S3 for converting into a directed graph based on the design CAD data and the system diagram data, a defect information acquisition step S4 for acquiring a past defect graph from a past piping design defect database 20, a candidate group generation step S5 for generating a candidate group by subgraph isomorphism of the directed graph for the acquired past defect graph, a similarity calculation step S6 for calculating the similarity between the candidate group and the past defect graph, a ranking presentation step S7 for arranging the candidate group and the past defect graph in order of similarity and presenting a ranking, and a design correction step S8.
In addition, since design CAD data is often managed as three-dimensional data and piping system diagram data is often managed as two-dimensional data, CAD data acquisition step S1 and system diagram data acquisition step S2 are separated. However, if the data are managed in an integrated manner in association with each other, it is also possible to integrate CAD data acquisition step S1 and system diagram data acquisition step S2.

In the CAD data acquisition step S1, design CAD data related to the design of the piping, such as 3D CAD or 2D drawings currently being designed, is acquired. The design CAD data to be acquired includes the following data A and data B.
Data A: Piping data expressed in line format for each pipe Data B: Center coordinate position and coordinate values of the smallest rectangular parallelepiped including the shape parts as shape parts Note that "piping data expressed in line format" is a representation of a single pipe with a line represented by a center line. The data for a single pipe is represented by a line segment with two points, as shown in the upper part of Figure 3(b). Also, "shape parts" are flanges, valves, equipment, etc. other than pipes. The shape parts are represented by bounding boxes, which are rectangular frames, as shown in the lower part of Figure 3(b).
In the system diagram data acquisition step S2, system diagram data of the piping is acquired. The acquired system diagram data includes the following data C.
Data C: A list of the names of the start and goal of the flow, and the goals reached from each start as a flow (see FIG. 3(e)).
The design CAD data includes piping data expressed in line format and the coordinate values of the smallest rectangular parallelepiped of the shape parts, and the system diagram data includes the start and goal conditions of the piping flow. In this way, by using the design CAD data and system diagram data that meet the necessary conditions for converting to a directed graph, the design CAD data and system diagram data can be converted to a directed graph with high accuracy.

In the directed graph conversion step S3, the design CAD data, which is non-graph structure data, is converted into a directed graph based on the design CAD data and system diagram data, so that the piping system is represented as a directed graph of nodes and edges.
FIG. 4 is a flow diagram of a method for converting to a directed graph.
The directed graph conversion step S3 includes a definition step S3-1 for defining graph nodes from line-format piping data and rectangular parallelepiped coordinate data, and defining edges connecting nodes from the line-format piping data, a contraction step S3-2 for contracting nodes, and a directed graph construction step S3-3 for searching for nodes based on start and goal conditions, and constructing a directed graph by assigning the shortest path and flow direction. This makes it possible to convert design CAD data and system diagram data into directed graphs efficiently and with high accuracy.

In the definition step S3-1, the nodes of the graph are defined from the above data A and data B. At this time, the position coordinates of both ends (start point and end point) of the pipe are defined as the attribute values of the node. In addition, the name of the node and the type of both ends are defined. For shape parts other than the pipe, such as flanges, the node is defined as the attribute value of the center position, and the name of the shape part is defined.
Additionally, in the definition step S3-1, edges connecting the nodes representing both ends of the piping line are defined from data A. At this time, the line of a single pipe is defined as an edge, and the length and the local cylindrical coordinate system (r, θ, Φ) are defined as attribute values. In addition, edges are defined between the nodes of other pipes and shape parts connected to the shape parts.

The contraction step S3-2 includes a first contraction step S3-2-1 for contracting the nodes of the point group represented by all the nodes, taking into consideration the range of the specified radius, and a second contraction step S3-2-2 for contracting the nodes of the shape parts and the other internal point nodes into one, taking into consideration other points inside the rectangular parallelepiped. By contracting, conversion to a directed graph can be performed appropriately.
In the first contraction step S3-2-1, for the group of points represented by all the nodes obtained in the definition step S3-1, the nodes existing within a specified radius R are regarded as one node, and these nodes are contracted.
In the second contraction step S3-2-2, if there is another point inside the rectangular parallelepiped in data B, the node representing the shape part in data B and the node existing inside are contracted into one node.
For example, in FIG. 3C, points within a distance of 5 mm and points inside a bounding box and the bounding box are contracted into one node.

In the directed graph construction step S3-3, the undirected graph constructed in the definition step S3-1 and contraction step S3-2 is searched for nodes corresponding to the start point and goal point of data C, and the shortest path connecting the nodes between those two points is searched for using the Dijktra algorithm. The flow direction is assigned to the order of the resulting paths, and a directed graph is constructed. Bidirectional flows are also permitted.
In this way, by converting the non-graph structure data of the design CAD into graph structure data, the input directed graph G _i is obtained as an abstract model that reflects the partition, system, route (position information), flow rate, distance to other components, etc. Fig. 3(d) shows a graph-format data structure with flow information as an example of this abstract model.

2A, a past defect graph is acquired from the past piping design defect database 20. In the past piping design defect database 20, defect information of past piping designs is stored as a past defect graph.
Here, a method for constructing the past piping design defect database 20 will be described.
The method for constructing the past piping design defect database 20 is a method for constructing past piping design defects as a database, and includes a design defect information acquisition step S11 for acquiring information on past piping design defects in order to obtain information as the fact that there was a defect in the past piping design, a past CAD data acquisition step S12 for acquiring past design CAD data related to the past piping design defects according to the information as the fact, a past system diagram data acquisition step S13 for acquiring past system diagram data related to the past piping design defects, a past design directed graph conversion step S14 for converting the past piping design defects into a directed graph based on the past design CAD data and the past system diagram data, and a storage step S15 for storing the directed graphed design defects as a past defect graph in the past piping design defect database 20. In this way, by constructing a database in which past design CAD data and past system diagram data that had a piping design defect in the past are appropriately converted into a directed graph and accumulated, a defect in the current piping design can be quickly found based on the similarity of the directed graph.

The information on past piping design defects in the design defect information acquisition step S11 and the past design CAD data in the past CAD data acquisition step S12 are acquired from a performance database 50 in which performance data on past piping designs is accumulated.
The performance database 50 records CAD data of each pipe system in each section for each ship. In the example shown in FIG. 1, for the drainage pipe system of section A in ship A, first review data, which is CAD data that has undergone a first design review, second review data, which is CAD data that has undergone a second design review, and completed CAD data, which is CAD data whose design has been completed, are recorded together with information on "defects" or "no defects". The first review data and second review data are assigned information on "defects", and the completed CAD data is assigned information on "no defects". The past design CAD data may be acquired from a CAD system linked to information on past design defects of piping. In addition, the information on design defects may include not only review but also defect information related to design acquired during manufacturing, trial operation, and service.

In the previous CAD data acquisition step S12, similarly to the CAD data acquisition step S1, the previous design CAD data to be acquired includes piping data expressed in line format and the coordinate values of the smallest rectangular parallelepiped of the shape parts.
In addition, in the past system diagram data acquisition step S13, similarly to the system diagram data acquisition step S2, the past system diagram data to be acquired includes start and finish conditions of the piping flow in the past.
The past design CAD data includes piping data expressed in line format and the coordinate values of the smallest rectangular parallelepiped of the shape parts, and the past system diagram data includes the start and goal conditions of the past piping flow. In this way, by using the past design CAD data and the past system diagram data that satisfy the necessary conditions for converting to a directed graph, the past design CAD data and the past system diagram data can be converted to a directed graph with high accuracy.

The previous design directed graph conversion step S14 is the same as the directed graph conversion step S3. That is, as shown in Fig. 4, the following steps are executed: definition step S3-1 in which the nodes of the graph are defined from the line-format piping data and the coordinate value data of a rectangular parallelepiped, and the edges connecting the nodes are defined from the line-format piping data; contraction step S3-2 in which the nodes are contracted; and directed graph construction step S3-3 in which the nodes are searched for based on the start and goal conditions, and the shortest path and the flow direction are assigned to construct a directed graph. This allows the previous design CAD data and system diagram data to be converted into a directed graph efficiently and with higher accuracy.
The contraction step S3-2 includes a first contraction step S3-2-1 for contracting the nodes of the point group represented by all the nodes, taking into consideration the range of the specified radius, and a second contraction step S3-2-2 for contracting the nodes of the shape parts and the other internal point nodes into one, taking into consideration other points inside the rectangular parallelepiped. By contracting, conversion to a directed graph can be performed appropriately.

In the directed graph construction step S3-3, the undirected graph constructed in the definition step S3-1 and contraction step S3-2 is searched for nodes corresponding to the start point and goal point of data C, and the shortest path connecting the nodes between those two points is searched for using the Dijktra algorithm. The flow direction is assigned to the order of the resulting paths, and a directed graph is constructed. Bidirectional flows are also permitted.
In this manner, the design CAD data that was defective in the past design is converted into a directed graph _Gf .

In the storage step S15, the graph of only the defect points in the directed graph _Gf is stored as a database of defect graphs {g ₁ , g ₂ , g ₃ , . . . }.
FIG. 5 is a diagram showing an example of data stored in the past piping design defect database.
The past piping design defect database 20 serves as defect dictionary data by accumulating past defect graphs such as those shown on the left side of Fig. 5. In addition, the past piping design defect database 20 also stores CAD data after past corrections and data on related documents (correction instructions, etc.) in association with defects, as shown on the right side of Fig. 5.
Using these past defect graphs as search keys, a similarity search is performed on the input directed graph G _i .

The steps from candidate generation step S5 to ranking presentation step S7 shown in FIG. 2A are a calculation procedure (similarity search method) for similarity search of past defect graphs _gj for an input directed graph _Gi .
First, in the candidate set generation step S5, a candidate set is generated by subgraph isomorphism.
If the input directed graph is expressed as _Gi = ( _Vi , _Ei ) and the past fault graph is expressed as _gj = ( _gVj , _gEj ) as a set of graph nodes and edges, then _gj is a subgraph isomorphism of _Gi if there exists a mapping f for any two vertices u, v of _gj such that the following equation (1) holds:
{u, v}∈E _i ⇒ {f(u), F(v)}∈gE _j ... (Math. 1)
The problem of determining the isomorphism of this subgraph is solved using the method of Ullman et al. (The design and analysis of computer algorithms, Alfred V. Aho, John E. Hopcroft, Jeffrey D. Ullman, Addison-Wesley Publishing, pp84-86,1974), and a candidate set {G _i,0 , G _i,1 , G _i,2 , ...} ∈ G _i corresponding to the past failure graph g _j is created.

Next, in a similarity calculation step S6, a rule-based similarity calculation is performed.
Between any candidate G _i,k = (V _i,k , E _i,k ) obtained in the candidate group generation step S5 and the past fault graph g _j = (gV _j , gE _j ), there is a mapping f from gV _j to V _i,k . In addition, since gV _j and gE _j have the attribute values defined in the definition step S3-1, a rule-based similarity calculation is performed using the mapping f and the attribute values according to the type of fault. This is performed for all candidate groups.

Next, in a ranking presentation step S7, a ranking format display is performed.
In the similarity calculation step S6, the similarity of the candidate group { _Gi,0 , Gi _,1 , Gi _,2 , ...} ∈ _Gi for the past fault graph _gj is obtained. Candidates with similarity equal to or greater than a specified threshold are sorted in descending order of similarity and presented in a ranked format. By obtaining past similar fault information ranked based on similarity, it is possible to quickly detect faults in the design of piping, such as pipes not being connected or flowing in the opposite direction at a pipe junction.
In addition, in the design correction step S8 shown in Fig. 2(a), the designer or the like corrects the design by referring to defects with high similarity from the similarity results presented in the ranking format. This allows the piping design defects to be appropriately corrected. The correction of the piping design defects can be performed manually or automatically by a computer, CAD system, or the like.

The above-mentioned method for detecting piping design defects can also be performed using a program.
A program for detecting piping design defects by referring to past defects causes a computer to execute a CAD data acquisition step S1, a system diagram data acquisition step S2, a directed graph conversion step S3, a defect information acquisition step S4, a candidate group generation step S5, a similarity calculation step S6, and a ranking presentation step S7 in a method for detecting piping design defects.
The program causes a computer to detect piping design defects by referring to past defects, thereby enabling piping design defects to be detected quickly and reducing labor.

Also, steps from candidate group generation step S5 to ranking presentation step S7 can be executed by AI (artificial intelligence). In this case, the similar defect search AI unit calculates the identity of the partial graph of the piping system and the rule-based or machine learning-based similarity from the past piping design defect database 20, and displays similar past defects in a ranked format according to the highest similarity, and also presents the appearance of the past defects after correction and document data related to the past defects.

The method for constructing the past piping design defect database 20 can also be performed using a program.
The program for constructing a database of past piping design defects causes a computer to execute a design defect information acquisition step S11, a past CAD data acquisition step S12, a past system diagram data acquisition step S13, a past design directed graph conversion step S14, and a storage step S15 in a method for constructing a past piping design defect database 20.
By having the program execute the construction of a database relating to past piping design defects on a computer, it is possible to reduce the labor required for constructing a database that accumulates past piping design defects.

FIG. 6 is a configuration diagram of a system for detecting defects in piping design.
The piping design defect detection system is a detection system that detects piping design defects by referring to past defects, and includes a design means 10 that designs piping using a CAD system 11, a past piping design defect database 20 that stores defect information of past piping designs as past defect graphs, and a defect detection means 30.
The design means 10 includes a CAD system 11 and a display unit 12 such as a monitor. A designer uses the CAD system 11 to design piping.
The past piping design defect database 20 is constructed based on a past piping design defect database construction method, and has a past design defect information acquisition unit 21, a past design directed graph conversion unit 22, a design defect storage unit 23, and a past defect memory unit 24.
The past design defect information acquisition unit 21 is used to execute the design defect information acquisition step S11, the past CAD data acquisition step S12, and the past system diagram data acquisition step S13.
The past-designed digraph conversion unit 22 is used to execute the past-designed digraph conversion step S14.
The design defect storage unit 23 is used to execute the storage step S15, and stores the past defect graph converted into a directed graph as an abstract model in the past defect memory unit 24. By constructing a database in which past design CAD data and past system diagram data in which there was a piping design defect in the past are appropriately converted into a directed graph and stored, defects in the current piping design can be quickly found based on the similarity of the directed graph.
Furthermore, the past piping design defect database 20 also stores corrected design CAD data and document data in association with piping systems that have had defects in the past.

The defect detection means 30 includes a design information acquisition unit 31 that acquires design CAD data and system diagram data of the piping designed by the design means 10, a past defect information acquisition unit 32 that acquires defect information of past piping designs from the past piping design defect database 20, a similarity search unit 33 that converts the design CAD data and the system diagram data into a directed graph, calculates the similarity with the defect information of the past piping designs, and calculates a ranking, and a result presentation unit 34 that presents the ranking.
The design information acquisition unit 31 is used to execute the CAD data acquisition step S1 and the system diagram data acquisition step S2.
The past defect information acquisition unit 32 is used to execute the defect information acquisition step S4.
The similarity search unit 33 includes a digraph conversion unit 33A, a candidate group generation unit 33B, a similarity calculation unit 33C, and a ranking calculation unit 33D. The digraph conversion unit 33A is used to execute the digraph conversion step S3, the candidate group generation unit 33B is used to execute the candidate group generation step S5, and the similarity calculation unit 33C is used to execute the similarity calculation step S6. The ranking calculation unit 33D and the result presentation unit 34 are used to execute the ranking presentation step S7. The ranking calculation unit 33D calculates a ranking based on the similarity calculated by the similarity calculation unit 33C. This allows the ranking calculation to be performed by accurately calculating the ranking of the similarity between the digraph of the current piping design and the digraph of the past piping design.
The result presentation unit 34 presents candidates with similarity equal to or greater than the specified threshold in order of similarity based on the calculated ranking. By obtaining past similar defect information ranked based on similarity, it is possible to quickly and accurately detect design defects in piping, such as pipes not being connected or a flow in the opposite direction at a pipe junction.
The display unit 12 of the design means 10 displays the rankings presented by the result presentation unit 34 together with related information. The related information is, for example, the appearance after the defect is corrected and document data. From the similarity results presented in the form of a ranking, a designer or the like can appropriately correct the design defect of the piping by referring to the defects with high similarity and the related information.

It is possible to install the design means 10, the past piping design defect database 20, and the defect detection means 30 all in the same location, but as shown in FIG. 6, the design means 10, the past piping design defect database 20, and the defect detection means 30 can be installed in separate locations, or, for example, the design means 10 and the past piping design defect database 20 can be installed in the same location and only the defect detection means 30 can be installed in a different location. In this way, the design means 10, the past piping design defect database 20, and the defect detection means 30 can be arbitrarily combined and information can be linked using the information communication network 40.
This enables one company to distribute the various means and databases across multiple locations, or to install the design means 10 and defect detection means 30 at a design company and only the past piping design defect database 20 at a database provider, making it possible for each company to use the system according to their respective business operations.

An example in which data is created that reproduces a defect that has been previously corrected by an expert and the present invention is applied will be described.
In this example, a piping defect where fluids passing through the pipes merge in the opposite directions was reproduced and stored in the past piping design defect database 20. As input, three reverse merging defects were artificially created in the drainage pipe system of one ship. In addition, a defect where two pipes have an overlapping portion was reproduced for the purpose of verifying the pipe connection check. This is difficult to judge by appearance alone. In addition, for reverse merging, the flow angle was taken as a parameter, and the reciprocal of the mean square of the difference was defined as the similarity.

FIG. 7 shows the results obtained in this example.
FIG. 7(a) shows a defect detected by the pipe connection check. FIG. 7(a) shows a pipe system viewed from above, where the rectangular parallelepiped X is the bounding box of the shape part, the dotted line represents the center line of the pipe on the upstream side of the flow, and the solid line represents the center line of the pipe on the downstream side of the flow. In FIG. 7(a), the pipes appear to be connected at first glance, but they overlap on the same straight line and are determined to be not directly connected. Naturally, conversely, when there is a gap, the pipes are also determined to be not connected. Note that this pipe connection check is performed in the procedure of constructing a directed graph, which is a stage before similarity search.

7(c) and 7(d) show the results of applying the present invention to the reverse junction pipe shown in FIG. 7(b) as the search target.
7C is an example of a pipe that was detected as the third highest degree of similarity. The solid line indicates the center line of the pipe, and the dotted line indicates the point where the degree of similarity was determined to be high. The pipe joins the main flow that flows from the upper left to the lower right in the opposite direction, and as a result of calculation by the similarity calculation unit 33C, the pipe was detected with a high degree of similarity of about 0.88.
On the other hand, FIG. 7D shows a result in which the similarity is 0.67, but this is not a defect because the merging is not in the reverse direction.
In this way, in this example, it was possible to detect locations close to past defect examples with a high degree of similarity.

The present invention can be used to check for defects in piping design in shipbuilding. It can also be expanded to wiring design and tube system design, in addition to piping design. It can also be expanded to other products such as floating bodies, offshore wind power generation facilities, underwater vehicles and marine structures, for which a similar analogy can be applied, as well as other industries such as the construction industry. When applied to these, the word "ship" in the claims can be replaced with words that apply to other products or other industries.

10 Design means 11 CAD system 20 Past piping design defect database 30 Defect detection means 31 Design information acquisition unit 32 Past defect information acquisition unit 33 Similarity search unit 33A Digraph conversion unit 33B Candidate group generation unit 33C Similarity calculation unit 33D Ranking calculation unit 34 Result presentation unit 40 Information communication network S1 CAD data acquisition step S2 System diagram data acquisition step S3 Digraph conversion step S3-1 Definition step S3-2 Reduction step S3-2-1 First reduction step S3-2-2 Second reduction step S3-3 Digraph construction step S4 Defect information acquisition step S5 Candidate group generation step S6 Similarity calculation step S7 Ranking presentation step S11 Design defect information acquisition step S12 Past CAD data acquisition step S13 Past system diagram data acquisition step S14 Past design digraph conversion step S15 Storage step

Claims (17)

A method for detecting design defects in piping by referring to past defects, comprising:
a CAD data acquisition step of acquiring design CAD data related to the design of the piping;
A system diagram data acquisition step of acquiring system diagram data of the piping;
a directed graph conversion step of converting the design CAD data and the system diagram data into a directed graph;
A defect information acquisition step of acquiring a past defect graph from a past piping design defect database;
a candidate set generation step of generating a candidate set based on a subgraph isomorphism of the directed graph for the acquired past failure graph;
a similarity calculation step of calculating a similarity between the candidate group and the past failure graph;
A method for detecting a fault in a piping design, comprising: a ranking presentation step of arranging the candidate group and the past fault graphs in order of similarity and presenting a ranking. The method for detecting piping design defects described in claim 1, characterized in that the design CAD data includes piping data expressed in line format and coordinate values of the smallest rectangular parallelepiped of the shape parts, and the system diagram data includes start and finish conditions of the flow of the piping. The piping design defect detection method according to claim 2, characterized in that the directed graph conversion step includes a definition step for defining graph nodes from the piping data in the line format and the coordinate value data of the rectangular parallelepiped, and defining edges connecting the nodes from the piping data expressed in the line format, a contraction step for contracting the nodes, and a directed graph construction step for searching for the nodes based on the start and goal conditions, and constructing the directed graph by assigning the shortest path and the flow direction. The method for detecting piping design defects according to claim 3, characterized in that the contraction step includes a first contraction step of contracting the nodes for the point cloud represented by all of the nodes, taking into account a specified radius range, and a second contraction step of contracting the nodes of the shape part and other internal point nodes into one, taking into account other points inside the rectangular parallelepiped. The method for detecting piping design defects according to any one of claims 1 to 4, characterized in that the design defect of the piping is corrected based on the past defect graph presented as the ranking in the ranking presentation step. A program for detecting piping design defects by referring to past defects, the program causing a computer to execute the CAD data acquisition step, the system diagram data acquisition step, the directed graph conversion step, the defect information acquisition step, the candidate group generation step, the similarity calculation step, and the ranking presentation step in the piping design defect detection method of any one of claims 1 to 4. A method for constructing a database of past piping design defects, comprising:
A design defect information acquisition step of acquiring information on the design defect of the past piping;
A past CAD data acquisition step of acquiring past design CAD data related to the design defect of the past piping;
A past system diagram data acquisition step of acquiring past system diagram data related to the design defect of the past piping;
a past design directed graph conversion step of converting the design defects of the past piping into a directed graph based on the past design CAD data and the past system diagram data;
A method for constructing a past piping design defect database, comprising: a storage step of storing the design defect converted into the directed graph as a past defect graph in a past piping design defect database. The method for constructing a past piping design defect database according to claim 7, characterized in that the past design CAD data includes piping data expressed in line format and coordinate values of the smallest rectangular parallelepiped of the shape parts, and the past system diagram data includes start and finish conditions of the past piping flow. The past design directed graph conversion step includes a definition step of defining graph nodes from the piping data in the line format and the coordinate value data of the rectangular parallelepiped, and defining edges connecting the nodes from the piping data expressed in the line format, a contraction step of contracting the nodes, and a directed graph construction step of searching for the nodes based on the start and goal conditions, and constructing a directed graph by assigning the shortest path and the flow direction. The past piping design defect database construction method according to claim 8. The method for constructing a past piping design defect database according to claim 9, characterized in that the contraction step includes a first contraction step for contracting the nodes in consideration of a specified radius range for the point cloud represented by all the nodes, and a second contraction step for contracting the nodes of the shape part and other internal point nodes into one in consideration of other points inside the rectangular parallelepiped. A program for constructing a database of past piping design defects, the program causing a computer to execute the design defect information acquisition step, the past CAD data acquisition step, the past system diagram data acquisition step, the past design directed graph conversion step, and the storage step in the method for constructing a past piping design defect database according to any one of claims 7 to 10. A detection system that detects piping design defects by referring to past defects, comprising: design means for designing the piping using a CAD system; a past piping design defect database that stores defect information of past piping designs as past defect graphs; a design information acquisition unit that acquires design CAD data and system diagram data of the piping designed by the design means; a past defect information acquisition unit that acquires the defect information of the past piping designs in the past piping design defect database; a similarity search unit that converts the design CAD data and the system diagram data into a directed graph and calculates a similarity with the defect information of the past piping designs to calculate a ranking; and a result presentation unit that presents the ranking. The piping design defect detection system according to claim 12, characterized in that the similarity search unit has a directed graph conversion unit, a candidate group generation unit, a similarity calculation unit, and a ranking calculation unit. The piping design defect detection system according to claim 12 or 13, characterized in that the past piping design defect database is constructed based on the past piping design defect database construction method according to any one of claims 7 to 10. The system for detecting piping design defects according to any one of claims 12 to 14, characterized in that the defect detection means is used in the method for detecting piping design defects according to any one of claims 1 to 5. The piping design defect detection system according to any one of claims 12 to 15, characterized in that the rankings of the result presentation unit are displayed on the design means together with related information. The piping design defect detection system according to any one of claims 12 to 16, characterized in that the design means, the past piping design defect database, and the defect detection means are arbitrarily combined and information is linked using an information communication network.