A Multi-objective Evolutionary Algorithm Based on Decomposition—Dynamic Resource Allocation with Mixture Model

Multi-objective evolutionary algorithm based on decomposition (MOEA/D) is the basis for studying the concept of multi-objective optimization problems on the basis of decomposition. MOEA/D has computational complexity in solving the multi-objective optimization sub-problems. To reduce computational complexity, in 2009, Zhang et al. proposed a method called MOEA/D-DRA (MOEA/D-Dynamic Resource Allocation). In MOEA/D-DRA, each sub-problem utilizes the computational resources dynamically in a different way. The construction of weight vectors is very important for the success of these algorithms. They actually define the search directions and distribution of the Pareto optimal solutions (POS). This paper proposes a weight vector generation method using uniform design and a mixture model (called MOEA/D-DRA-MM). The MOEA/D-DRA-MM builds a new weight vector generation method that makes the distribution of the weights uniform and broadens the diversity and convergence. MOEA/D-DRA-MM uses two weight vectors that are obtained by uniform design with a mixture distribution. Simulation results suggest that MOEA/D-DRA-MM may provide better convergence and diversity with respect to inverted generational distance and hypervolume metrics.