Nested GA for Multi-objective Bilevel Optimization A Case of Mixed-model Assembly Line Planning

A multi-objective optimization problem (MOP) is a decision problem for two or more conflicting goals. Traditional solutions generally use sequential optimization or All-in-one methods. The sequential optimization method only finds the optimal solution for a single problem, and it is difficult to obtain an overall optimality. The All-in-one approach simplifies a composite conflicting objective function in the form of a weighted sum. This kind of weighted solution based on preference weights is highly subjective and tends to sacrifice some goals. The goal of MOP is to coordinate and combine all goals. A multi-objective bilevel optimization model is proposed in this paper. A coordination model between two interconnected objectives is established by a two-level hierarchical optimization mechanism. Aiming at the bi-level programming model, a bi-level, nested genetic algorithm and its corresponding encoding strategy are developed.