Pattern Nesting Using Clustering and Genetic Algorithm
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing pattern nesting methods struggle to efficiently arrange multiple patterns on a fabric while minimizing fabric waste, especially when dealing with a large number of patterns.
Innovation Solution
A pattern nesting method using clustering and a genetic algorithm, where patterns are grouped by size and arranged on the fabric based on predetermined positional relationships and optimization techniques, including preprocessing to reduce mesh complexity and using a GPU for faster computations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of substance
If traditional pattern nesting methods are used to arrange patterns on fabric, then the arrangement process becomes manageable, but fabric waste increases and the number of patterns that can be efficiently nested is limited
Solution Approach 1:
The patent divides the pattern nesting problem into two main segments: (1) generating a pattern pair that determines positional relationships between patterns, and (2) nesting patterns into fabric using the pattern pair and genetic algorithm. This segmentation allows the system to handle large numbers of patterns efficiently by breaking down the complex arrangement task into manageable components.
Solution Approach 2:
The patent performs preliminary actions by generating a pattern pair that establishes predetermined positional relationships between patterns before the actual nesting process. This preliminary determination of relative positions (such as which pattern is above, below, left, or right of another) enables more efficient fabric utilization during the subsequent nesting phase, reducing fabric waste.
2Productivity
If the number of patterns to be nested increases, then production capacity improves, but the complexity of arranging patterns without overlap and maximizing fabric utilization increases
Solution Approach 1:
The patent employs a genetic algorithm that dynamically adjusts and evolves pattern arrangements through iterative optimization. The system generates candidate solutions, evaluates them based on fabric utilization and overlap constraints, and evolves better arrangements over generations. This dynamic approach enables handling of large numbers of patterns by adapting the arrangement strategy rather than relying on static rules.
Solution Approach 2:
The genetic algorithm incorporates feedback mechanisms by evaluating arrangement quality based on overlap detection and fabric utilization metrics. The system uses this feedback to guide the evolution of pattern arrangements, selecting and refining configurations that minimize waste and satisfy constraints, thereby managing complexity even as the number of patterns increases.
3Loss of substance
If patterns are arranged to maximize fabric utilization, then fabric waste reduces, but the computational time and complexity of finding optimal arrangements increases
Solution Approach 1:
By pre-generating a pattern pair that establishes relative positional relationships, the patent reduces the search space for optimal arrangements. This preliminary structuring constrains the possible configurations, allowing the genetic algorithm to focus computational effort on optimizing within a reduced space, thereby reducing computational time while still achieving high fabric utilization.
Solution Approach 2:
The patent changes the parameter representation by using relative positional relationships (pattern pairs) rather than absolute coordinates for all patterns. This parameter transformation simplifies the optimization problem by reducing the number of variables and constraints that need to be evaluated, enabling faster computation of near-optimal arrangements that minimize fabric waste.
Data Source
AI summary
A pattern nesting method according to one embodiment may comprise the steps of: generating a pattern pair in which the location relationship between at least two patterns is determined so that relative locations of at least two patterns from among a plurality of patterns satisfy predetermined conditions; clustering the plurality of patterns on the basis of size information about each of the plurality of patterns; and nesting, on the basis of the pattern pair and a genetic algorithm, onto a fabric, at least one pattern included in each of the plurality of clusters generated on the basis of the clustering result.