Iterative Pattern Generation for Steel Plate Cutting Layouts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In steel plants, efficiently cutting large steel plates into smaller plates of various sizes is challenging due to the difficulty in calculating layouts that minimize waste and maximize yield, especially when dealing with atypical sizes and elements of different dimensions.
Innovation Solution
A computer-implemented method and apparatus that selects candidate elements for grouping based on evaluation values, determines whether to generate groups, and weights element evaluation values based on their inclusion in already-generated groups to optimize layout patterns for cutting, ensuring efficient use of material without duplication.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional layout calculation methods are used for cutting steel plates, then the calculation process becomes complex and time-consuming, but the yield optimization is insufficient
Solution Approach 1:
The patent segments the cutting problem into discrete elements (steel plates to be cut) and groups them by size categories. The layout calculation is divided into multiple iterations where patterns are generated step-by-step, with evaluation values calculated for each grouping. This segmentation transforms the complex continuous optimization problem into discrete, manageable steps that can be processed iteratively.
Solution Approach 2:
The patent performs preliminary actions by pre-defining size categories for steel plates and pre-establishing grouping patterns before the actual layout optimization. Elements are pre-classified into size groups, and evaluation functions are pre-configured to assess different grouping patterns. This preliminary structuring reduces the complexity of the subsequent optimization process.
2Productivity
If multiple layout patterns are generated to improve yield, then the calculation time increases significantly
Solution Approach 1:
The patent implements periodic action through iterative pattern generation, where a predetermined number of patterns are generated in each iteration cycle. The evaluation values are calculated periodically for each grouping pattern, and the process repeats for multiple iterations. This periodic approach allows the system to explore multiple layout possibilities while maintaining a structured timeline for calculation, balancing comprehensiveness with time management.
Solution Approach 2:
The patent incorporates feedback mechanisms where evaluation values are calculated for each generated pattern based on yield efficiency. The system uses this feedback to determine whether to continue generating patterns or to accept the current best solution. The iterative process adjusts pattern generation based on previous evaluation results, allowing the calculation to converge on optimal solutions without exhaustively exploring all possible patterns.
3Manufacturing precision
If strict grouping constraints are applied to avoid duplication, then the flexibility in layout design is reduced
Solution Approach 1:
The patent applies dynamics by making the grouping constraints adaptive rather than static. The evaluation function dynamically assesses whether adding an element to a group violates size constraints, and the iterative process allows constraints to be re-evaluated in each iteration. This dynamic approach maintains manufacturing precision (no duplication or size violations) while preserving layout flexibility through multiple possible pattern configurations that can be explored over iterations.
Data Source
AI summary
A method to generate a plurality of groups each including at least one of a plurality of elements. The method includes selecting at least one candidate element from the plurality of elements as a candidate to be included in a group, determining whether or not to generate the group including the at least one candidate element selected in the selecting step, based on an element evaluation value associated with each of the at least one candidate element selected in the selecting step, provided that a determination is made to generate the group in the determining step, generating the group including the at least one candidate element selected in the selecting step, and weighting the element evaluation value of each of the at least one element according to how many times the each element is included in already-generated groups to reflect the weighted element evaluation value in next group generation.