Combining model-based and genetics-based offspring generation for multi-objective optimization using a convergence criterion

摘要

In a method for multi-objective optimization (MOO), a computer executes the following steps:
- Providing an initial population of solutions (310);
- Partitioning the population into a number of clusters (320);
- Calculating a convergence ratio (¨) for each cluster (330);
- Checking whether the convergence ratio (¨) is larger than a given threshold value (Á) (340),
- If the convergence ratio (È) is smaller or equal than the given threshold value (Á), generating model consisting of a deterministic part and a stochastic part. Generating offspring by sampling the model (350), else generating offspring using mutation and crossover (360);
- Merging parents and offspring (370); and
- Creating a new parent population using a Pareto-based selection selection method (370).