Evolutionary Algorithm Variable Constraint for Feasible Pareto Solutions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Multi-objective optimization in product design using evolutionary algorithms often results in infeasible combinations of explanatory variables, as the algorithms aim to calculate Pareto solutions with higher fitness degrees, leading to unachievable outcomes.
Innovation Solution
An information-processing method and device that generate new individuals by altering explanatory variables based on predefined probabilities and values from current generation data, ensuring feasible solutions by selecting individuals based on fitness degrees and applying constraint conditions to continuous or discrete variables.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If evolutionary algorithms are used to calculate Pareto solutions with higher fitness degrees, then optimization performance is improved, but feasibility of solutions deteriorates
Solution Approach 1:
The patent applies preliminary action by pre-defining feasible ranges and combination conditions for explanatory variables before the optimization process begins. This ensures that all generated individuals automatically satisfy feasibility requirements while the algorithm searches for high-fitness solutions. The constraint conditions are established in advance to guide the evolutionary search toward feasible regions of the solution space.
Solution Approach 2:
The patent employs parameter changes by dynamically adjusting the values of explanatory variables within predefined feasible ranges during the evolutionary process. By modifying parameters such as continuous variable bounds and discrete variable options according to combination conditions, the algorithm maintains solution feasibility while exploring different regions of the search space to improve fitness.
2Manufacturing precision
If combination conditions are applied to explanatory variables, then solution feasibility is improved, but algorithm complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the set of explanatory variables into different groups based on their combination conditions. Each group is handled with specific feasible ranges and constraints, allowing the algorithm to manage complexity through structured organization. This segmentation enables independent handling of different variable types while maintaining overall solution feasibility.
Solution Approach 2:
The patent employs universality by creating a unified framework that handles both continuous and discrete explanatory variables under a common combination condition system. The same basic mechanism applies to different variable types, reducing algorithmic complexity through standardized processing while maintaining the ability to handle diverse variable constraints.
Data Source
AI summary
An information-processing method executed by an information-processing device includes: generating a new individual by changing a value of part of explanatory variables of an individual selected with a first probability from current generation data, which has n individuals (n is an integer of three or more) and in which each individual is an explanatory variable group having a plurality of explanatory variables, to a target value based on a value of a corresponding explanatory variable of another individual; generating a new individual by changing a value of part of explanatory variables of an individual selected with a second probability from the current generation data to another value; and selecting n individuals of next generation data from the n individuals and the generated individual based on a fitness degree calculated from an explanatory variable group for each individual, wherein an explanatory variable having a value defined by a combination condition is set to take a value represented by one of a plurality of combinations constituted of a plurality of values determined in advance.