Visualization method for high-dimensional multi-objective optimization

Abstract

The invention discloses a visualization method for high-dimensional multi-objective optimization, and the method comprises the steps: giving a group of reasonable objective sequences through the analysis of all objectives in an optimal solution set obtained through various multi-objective optimization algorithms; for an optimal solution set obtained by a multi-objective optimization algorithm on ahigh-dimensional problem, carrying out non-dominated sorting to filter out dominated solution sets in the solution set so as to ensure that all optimal solutions are non-dominated solutions; then, through a kernel method, evaluating the association degree between the targets in the solution set, and generating a group of association matrixes; finding out a group of optimal sequences capable of visually reflecting conflicts and relations of all targets in the optimal solution set through the generated incidence matrix; and finally, displaying the reordered optimal solution set of the target onthe parallel coordinate axis. According to the method, a novel visualization method can be provided for optimal solutions generated by different optimization algorithms in an actual high-dimensionalmulti-objective problem, so that high-quality solution set information is provided for a decision maker, and finally, the time and the labor cost in the optimal solution analysis process are saved.

Description

technical field [0001] The invention relates to a visualization method for high-dimensional multi-objective optimization, which belongs to the technical field of computer software development. Background technique [0002] With the development of today's society, the field of artificial intelligence has attracted more and more public attention, and the multi-objective optimization problem for high dimensions has also begun to attract extensive attention from researchers at home and abroad. In actual production and engineering, high-dimensional multi-objective optimization problems often mean that the number of objective functions that the algorithm needs to optimize at the same time exceeds three, and the number of objectives of the optimal solution set obtained by the final algorithm will also be greater than three. In low-dimensional (two-dimensional, three-dimensional) optimization problems, the visualization of the final solution set is usually represented by a scatter p...

AI Technical Summary

Problems solved by technology

However, in high-dimensional multi-objective optimization problems, since the number of objectives in the solution set is greater than three, two-dimensional and three-dimensional scatter plots can no longer be used in the visualization of solution sets, and are replaced by high-dimensional parallel axis plots.

The high-dimensional parallel coordinate axis diagram can reflect the nature of the solution set to a certain extent, but because it can only display the solution set on the coordinate axis according to the order specified at the beginning of the goal, it will affect the relationship between the goals in the solution set to a certain extent. display of relationships

Images