KR-20260063220-A - METHOD FOR DETERMINING WEIGHTED VALUE OF OBJECTIVE FUNCTION IN AUTOMATIC DESIGN ALGORITHM OF PIPING ROUTE

KR20260063220AKR 20260063220 AKR20260063220 AKR 20260063220AKR-20260063220-A

Abstract

A method for determining the weight of an objective function in an automatic pipe path design model according to the present invention comprises: a pipe data input step for inputting data for a pipe to be designed into the automatic pipe path design model; a design area data input step for inputting data for an area for designing a pipe into the automatic pipe path design model; an objective function input step for inputting an objective function including a plurality of design variables and arbitrary weights for each design variable into the automatic pipe path design model; a pipe path generation step for generating a pipe path using the automatic pipe path design model; a matrix/vector representation step for representing the pipe path as a matrix or vector; a reference matrix/vector representation step for representing a predetermined reference pipe path as a matrix or vector; a difference calculation step for calculating the difference between two matrices or vectors; and a judgment step for determining whether the difference is less than or equal to a predetermined standard, wherein if the difference is less than or equal to the predetermined standard, the weight is determined as the final weight, and if the difference exceeds the predetermined standard, the weight of the objective function is adjusted, and then the objective function input step, the pipe path generation step, the matrix/vector representation step, the difference calculation step, and the judgment step are repeated.

Inventors

김정연
한인수
최성원
박상민
김성희
김미진
유한준
이현승
이인석
노명일
공민철

Assignees

에이치디한국조선해양 주식회사
에이치디현대중공업 주식회사
서울대학교산학협력단
에이치디현대삼호 주식회사

Dates

Publication Date: 20260507
Application Date: 20241030

Claims (8)

Piping data input step for inputting data for the piping to be designed into the automatic piping path design model; A design area data input step for inputting data for an area to design piping into the above-mentioned automatic piping path design model; An objective function input step of inputting an objective function including a plurality of design variables and arbitrary weights for each design variable into the above-mentioned automatic piping path design model; A pipe path generation step that generates a pipe path using the above pipe path automatic design model; A matrix/vector representation step for expressing the above piping path as a matrix or vector; Reference matrix/vector representation step representing a predetermined reference piping path as a matrix or vector; A difference calculation step for calculating the difference between two matrices or vectors; and Including a judgment step for determining whether the above difference is below a predetermined standard, If the difference is less than or equal to a predetermined standard, the weight is determined as the final weight, and if the difference exceeds a predetermined standard, the weight of the objective function is adjusted, and then the objective function input step, the pipe path generation step, the matrix/vector representation step, the difference calculation step, and the judgment step are repeated. Method for determining the weights of the objective function in an automatic piping path design model.
In paragraph 1, The design variables of the above objective function include two or more of minimizing the total length of the piping, minimizing the number of bends in the piping, maximizing the space utilization of the design area, minimizing the length of diagonal sections of the piping, minimizing piping material costs, and minimizing piping installation costs. Method for determining the weights of the objective function in an automatic piping path design model.
In paragraph 1, In the above matrix/vector representation step and the above reference matrix/vector representation step, the graph of each pipe path is represented as an adjacency matrix, Method for determining the weights of the objective function in an automatic piping path design model.
In paragraph 3, In the above difference calculation step, the difference is calculated using the Frobenius Norm, Method for determining the weights of the objective function in an automatic piping path design model.
In paragraph 1, In the above matrix/vector representation step and the above reference matrix/vector representation step, the graph of each pipe path is converted into an adjacency matrix and then vectorized to be represented as a vector, Method for determining the weights of the objective function in an automatic piping path design model.
In paragraph 1, In the above matrix/vector representation step and the above reference matrix/vector representation step, an embedding model is applied to the graph of each pipe path to represent it as a vector, Method for determining the weights of the objective function in an automatic piping path design model.
In paragraph 5 or 6, In the above difference calculation step, the difference is calculated through cosine similarity, Method for determining the weights of the objective function in an automatic piping path design model.
In paragraph 1, If the above difference exceeds a predetermined threshold, the weights of the above objective function are adjusted so that the difference can be reduced by applying a genetic model capable of multivariate optimization, the Simplex Method, Simulated Annealing, or an Evolution Strategy. Method for determining the weights of the objective function in an automatic piping path design model.

Description

Method for Determining Weighted Value of Objective Function in Automatic Piping Route Design Model The present invention relates to a method for determining the weights of an objective function in an automatic piping path design model. For example, as disclosed in published patent 10-2023-0166935 and registered patent 10-2386084, programs that automatically design piping paths are used when designing piping in ships or plants. These programs typically operate by inputting an objective function that includes multiple design variables and weights for each design variable, along with information about the piping to be designed and the area to be designed. The program then searches for a path that optimizes the objective function on a 3D grid composed of possible piping paths and automatically designs that path. In this case, since the program proposes a 'path that optimizes the objective function,' the resulting path varies depending on how the objective function is defined—for instance, how weights are assigned. Therefore, the objective function must be defined so that the resulting path closely approximates the actual design; however, currently, this is determined experimentally by the designer. Consequently, defining the objective function not only requires significant time and effort but also presents a problem where it can be heavily influenced by the designer's personal know-how or proficiency. FIG. 1 is a flowchart of a method for determining the weight of an objective function in an automatic piping path design model according to an embodiment of the present invention. A method for determining the weight of an objective function in an automatic piping path design model according to an embodiment of the present invention is described in detail with reference to the drawings. FIG. 1 is a flowchart of a method for determining the weight of an objective function in an automatic piping path design model according to an embodiment of the present invention. Referring to FIG. 1, the present method includes a pipe data input step (S1), a design area data input step (S2), an objective function input step (S3), a pipe path generation step (S4), a matrix/vector representation step (S5), a reference matrix/vector representation step (S6), a difference calculation step (S7), and a judgment step (S8). In the pipe data input step (S1), data for the pipe to be designed is entered into the pipe path automatic design model. Here, the automatic pipe path design model may be a commercial model, for example, a deep learning model configured based on various networks and trained to predict pipe paths using data including pipes, data not including pipes, and an objective function. Data for the piping to be designed may be provided, for example, as a P&ID (Piping and Instrumentation Diagram) or a piping list, and may include information regarding the start and end points of the piping, the diameter of the piping, the material of the piping, etc. In the design area data input step (S2), data for the area to be designed for piping is entered into the automatic piping path design model. Data regarding the area for designing piping may be, for example, a graph or drawing including the floor, walls, ceiling, various facilities or systems excluding piping, and other obstacles of the area. In the objective function input step (S3), the objective function is input into the pipe path automatic design model. The objective function includes multiple design variables and weights for each design variable. Design variables may include, for example, minimizing the total length of the piping, minimizing the number of bends in the piping, maximizing the space utilization of the design area, minimizing the length of diagonal sections of the piping, minimizing material costs, and minimizing installation costs. Here, minimizing the total length of the piping means that the piping route is designed to minimize the total length of the piping. In addition, minimizing the number of bends in the piping means that the piping path is designed so that the number of points where the direction of the piping changes is minimized. In addition, maximizing the space utilization of the design area means that the piping path is designed so that the space utilization of the design area is maximized when the piping is installed by extending the piping along the walls of the design area as much as possible or positioning it close to the walls of the design area. Minimizing the length of diagonal sections in a pipe means that the pipe path is designed to minimize the length of sections that extend diagonally in the pipe, such as 45° or 60°. Minimizing piping material costs means that the piping route is designed to minimize material costs. Minimizing piping installation costs means that the piping route is designed to minimize installation costs. However, the present invention is not limited to the design variables mentioned above, and other design variables may be