Selecting Optimal Solution from Pareto
Non-inferior Solutions
Introduction
After carefully observing the distribution of the Pareto front, we find that the
distribution is monotonically increasing or decreasing. This means that
different variability exists in the Pareto front and that new inherent disciplines
can be found.
The methods for solving Multi-objective Optimization Problems (MOP) can be
divided into three categories:
Functional Relation Method: Seek a functional relation between
objectives to convert to single objective problem.
Non-dominant Relation Method: Obtain Pareto solution by non-
dominant relations.
Evaluation Factors Method: Add evaluation system according to
preferences.
Some common methods include:
TOPSIS: Technique for Order Preference by Similarity to Ideal Solution.
Fuzzy Logic
Unsupervised ML: Automatically extract mathematical relations
All the methods have one thing in common, that is introduction of additional
conditions on the basis of optimization objectives. The key to solving MOPs lie
in directing the gains and losses properly.
To select the optimal solution, the rules are as follows:
Maximize Income Ratio (Performance/Price)
Assign priority order for objectives
The method is based on performance-price ratio and is only verified for a bi-
objective problem.
- 1
- 2
前往页